CN109614273A

CN109614273A - The read method and system of incremental data

Info

Publication number: CN109614273A
Application number: CN201811591294.4A
Authority: CN
Inventors: 郭勇; 陈鹏; 王子骏
Original assignee: GUANGZHOU DINGJIA COMPUTER TECHNOLOGY Co Ltd
Current assignee: GUANGZHOU DINGJIA COMPUTER TECHNOLOGY Co Ltd
Priority date: 2018-12-20
Filing date: 2018-12-20
Publication date: 2019-04-12
Anticipated expiration: 2038-12-20
Also published as: CN109614273B

Abstract

The present invention relates to the read method of incremental data and systems, belong to network technique field.The described method includes: obtaining incremental metadata；The incremental metadata is for describing incremental data；Obtain the write state label of the incremental metadata；The write state label is for characterizing whether corresponding incremental data is successfully written in disk；When the write state is labeled as label is write complete, corresponding incremental data is read from disk according to the incremental metadata.Above-mentioned technical proposal solves the problems, such as that mistake occurs in the reading of incremental data.It is marked by setting write state, can characterize whether corresponding incremental data is successfully written in disk, therefore target backup machine can be made to read correct incremental data from disk.

Description

The read method and system of incremental data

Technical field

The present invention relates to network technique field, more particularly to the read method of incremental data, system, computer equipment and Storage medium.

Background technique

Continuous data protection (CDP, Continual Data Protection) is a kind of based on the real-time of disk block device Redundancy technique.IO (Input/Output, input and output) request is encapsulated as monitoring record and is sent to backup by backup target function Server.Different from timed backup method, CDP redundancy technique is IO grades or microsecond grade to the protection granularity of data. In addition, can farthest reduce number when breaking down or when disaster, user can accurately select time point to carry out data recovery According to loss (RPO, Recovery Point Object).

In realizing process of the present invention, at least there are the following problems in the prior art for inventor's discovery: for using asynchronous Mode carries out the CDP redundancy technique of incremental data reading and transmission, which first constructs monitoring record and be submitted to this earth magnetism Then disk is sent in backup server from extraction monitoring record in disk and is backed up.Same position is being synchronized on disk Multiple IO write-in may occur before to backup server, and then cause real in the incremental data to be read or be written and disk The incremental data on border is in inconsistent state.There is mistake in this reading that this may result in incremental data.

Summary of the invention

Based on this, the embodiment of the invention provides the read method of incremental data, system, computer equipment and storages to be situated between Matter can improve the reading accuracy of incremental data.

The content of the embodiment of the present invention is as follows:

A kind of read method of incremental data, comprising the following steps: obtain incremental metadata；The incremental metadata is used for Incremental data is described；Obtain the write state label of the incremental metadata；The write state label is for characterizing corresponding increment Whether data are successfully written in disk；When the write state labeled as write complete label when, according to the incremental metadata from Corresponding incremental data is read in disk.

In one of the embodiments, before the step of write state label for obtaining the incremental metadata, also wrap It includes: receiving write IO request；It include disk offset in the write IO request；It is deviated and is determined according to the disk in the write IO request Corresponding incremental data is written in said write address by the writing address of disk.

In one of the embodiments, after the step of acquisition incremental metadata, further includes: by the increment member number According in the increment record queue being added in memory；It is described when the write state labeled as write complete label when, according to institute State the step of incremental metadata reads corresponding incremental data from disk, comprising: extract institute from the increment record queue State incremental metadata；It is marked according to extracted incremental metadata and the write state, corresponding increment is read from disk Data.

The write state label includes label to be done in one of the embodiments,；It is described according to extracted increment The step of metadata and the write state mark, corresponding incremental data are read from disk, comprising: when the increment member number When according to corresponding write state labeled as label is write complete, corresponding increment is read from disk according to extracted incremental metadata Data；When the corresponding write state of the incremental metadata is labeled as label to be done, judge whether next read cycle arrives It reaches；If so, reading corresponding incremental data from disk according to extracted incremental metadata.

The incremental metadata includes at least one of the following: disk offset and data length in one of the embodiments,； Before the step incremental metadata being added in the increment record queue in memory, further includes: by increment On the corresponding sector position of metadata write-in variation sector bitmap；The corresponding sector position is according to disk offset and data What length determined；When detecting that the variation sector bitmap, will be corresponding with overlapping sector position there are when the position of overlapping sector The incremental metadata being newly written is added in conflict record queue；Wherein, the overlapping sector position is new in variation sector bitmap The incremental metadata of the write-in sector position be overlapped with history incremental metadata.

It is described after the step of reading corresponding incremental data in disk in one of the embodiments, further includes: will Corresponding incremental metadata is removed in the overlapping sector position and conflict record queue；Conflict in traversal conflict record queue Incremental metadata, and judge in the bitmap of the variation sector with the presence or absence of overlapping sector corresponding with the conflict incremental metadata Position；When it be not present, determine that the conflict incremental metadata is read in advance；The conflict incremental metadata is added to different It causes in record queue；Wherein, the inconsistency record queue is used to carry out consistency before reading incremental data in disk It checks；The consistency check includes determining to need the consistent of the incremental data that reads and the incremental data being written in disk Property.

Described the step of corresponding incremental data is read from disk in one of the embodiments, comprising: read currently It takes in the period, when the number of the incremental metadata is more than given threshold, incremental data is read in stopping from disk, under One read cycle reaches.

Correspondingly, the embodiment of the present invention provides a kind of reading system of incremental data, comprising: filtration drive module is used for Obtain incremental metadata；Wherein, the incremental metadata is for describing incremental data；Label obtains module, receives for working as When data read request, the write state label of incremental metadata is obtained；The write state label is for characterizing corresponding increment member Whether data are successfully written in disk；And read module, for when the write state labeled as write complete label when, root Corresponding incremental data is read from disk according to the incremental metadata.

The read method and system of above-mentioned incremental data are marked by setting write state, can characterize corresponding incremental data Whether it is successfully written in disk, therefore target backup machine can be made to read correct incremental data from disk.

A kind of computer equipment can be run on a memory and on a processor including memory, processor and storage Computer program, the processor perform the steps of acquisition incremental metadata when executing the computer program；The increment Metadata is for describing incremental data；Obtain the write state label of the incremental metadata；The write state label is for characterizing Whether corresponding incremental data is successfully written in disk；When the write state is labeled as label is write complete, according to the increasing Amount metadata reads corresponding incremental data from disk.

Above-mentioned computer equipment is marked by setting write state, can characterize whether corresponding incremental data is successfully written to In disk, therefore target backup machine can be made to read correct incremental data from disk.

A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor Acquisition incremental metadata is performed the steps of when row；The incremental metadata is for describing incremental data；Obtain the increment member The write state of data marks；The write state label is for characterizing whether corresponding incremental data is successfully written in disk；When The write state reads corresponding incremental data according to the incremental metadata labeled as when writing complete label from disk.

Above-mentioned computer readable storage medium, by setting write state mark, can characterize corresponding incremental data whether by It is successfully written in disk, therefore target backup machine can be made to read correct incremental data from disk.

Detailed description of the invention

Fig. 1 is the applied environment figure of the read method of incremental data in one embodiment；

Fig. 2 is the flow diagram of the continuous data protection method without local cache in one embodiment；

Fig. 3 is the operation principle schematic diagram of filter drive program in one embodiment；

Fig. 4 is the formation basic theory schematic diagram of conflict record queue and inconsistency record queue in one embodiment；

Fig. 5 is the flow diagram of the read method of incremental data in one embodiment；

Fig. 6 is the flow diagram of the read method of data in one embodiment；

Fig. 7 is the flow diagram of the continuous data protection method without local cache in another embodiment；

Fig. 8 is the increment record conflict schematic illustration for putting inspection consistent with data in one embodiment；

Fig. 9 is that increment record continuity illustrates schematic diagram in one embodiment；

Figure 10 is the structural block diagram of the continuous data protection system without local cache in one embodiment；

Figure 11 is the structural block diagram of the reading system of incremental data in one embodiment；

Figure 12 is the structural block diagram of the reading system of data in one embodiment；

Figure 13 is the structural block diagram of the continuous data protection system without local cache in another embodiment；

Figure 14 is the internal structure of computer equipment in one embodiment.

Specific embodiment

In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and It is not used in the restriction present invention.

The continuous data protection method of no local cache provided by the present application can be applied to application environment as shown in Figure 1 In.It include backup target machine 101 and backup server 102, backup target machine 101 and backup server 102 in the application environment It can be communicated by network.Target backup machine 101 refers to the equipment for needing to carry out data backup, backup server 102 It is the server that the data on target backup machine 101 are backed up.Backup target machine 101 obtains incremental metadata, according to increasing Amount metadata reads incremental data from disk, and is sent to backup server 102 and is backed up.Wherein, backup target machine 101 It can be the various terminals for being stored with data, PC (personal computer), server etc..Server (including backup services Device 102) it can be realized with the server cluster of the either multiple server compositions of independent server.Certainly, standby to target What the data of part machine 101 were backed up realizes except through backup server, can also be by other with the hard of backup functionality Part equipment or virtual unit are realized.

The embodiment of the present invention provides the continuous data protection method of no local cache a kind of, system, computer equipment and deposits Storage media.It is described in detail separately below.

In one embodiment, as shown in Fig. 2, providing a kind of continuous data protection method of no local cache.With this Method is applied to be illustrated for the target backup generator terminal in Fig. 1, comprising the following steps:

S201, data backup requests are obtained；Incremental metadata is obtained according to the data backup requests；Wherein, the increasing Amount metadata is for describing incremental data.

Wherein, under the application environment of continuous data protection (hereinafter referred to as CDP), data backup requests can be according to writing IO Request triggering.When target backup machine intercepts the write IO request on upper layer, show that the data on target backup machine will do it write-in behaviour Make and the data in disk can change, it is necessary to backup to the data of variation in backup server.Therefore target backup Machine can convert write IO request to the data backup requests to corresponding incremental data, to trigger continuous data protection.

CDP has an initialization procedure, this can be all synchronized to the data of monitoring panel in backup server in the process, Enter real time monitoring process after the completion of initialization, write IO request response each time is synchronized compared to initialization during real time monitoring All it is an increment, therefore, the data being written in target backup machine can be known as incremental data.Further, IO is write It may include the description information of incremental data in request, such as: disk offset, data length, timestamp etc., these information can be with Referred to as metadata.It should be noted that may have muti-piece disk on backup target machine, target backup machine can be intercepted and captured only and is directed to The write IO request of a certain piece or a few pieces disk, these intercepted disks just become monitoring panel, i.e. monitoring panel refers to needing Carry out the disk of data backup.

The embodiment of the present invention is described by incremental metadata (be referred to as incremental metadata record or increment records) Incremental data may include the metadata such as the corresponding disk offset of incremental data, data length, timestamp letter in incremental metadata Breath.It should be noted that incremental metadata is for describing incremental data, it can not include incremental data itself.

Further, the acquisition of data backup requests can be realized by Kernel Filtering driver, and Kernel Filtering drives Dynamic program intercepts and captures the write IO request of upper layer transmission in real time, and is converted into the data backup requests to corresponding incremental data, and according to Disk offset, data length, timestamp in the data backup requests intercepted and captured etc. generate corresponding incremental metadata.

S202, corresponding incremental data is read from disk according to the incremental metadata.

Incremental data can store in the disc cache of disk, and when needing to carry out data backup by incremental data from magnetic It is read out in disk.Disc cache is disk of the target backup machine for the temporary data intercepted, it can in monitoring panel Incremental data carries out temporary disk.The reading of incremental data and incremental metadata can read routine by kernel to realize. Reading data work i.e. in target backup machine can read routine by kernel to realize.

S203, read incremental data is sent to backup server, to trigger the backup server to the increasing Amount data are backed up.

Target backup machine is written and read the data in monitoring panel according to write IO request, while target backup machine will be corresponding Incremental data is sent to backup server in real time, to realize continuous data protection, accurately time point can be selected to be counted According to recovery, data degradation can be farthest reduced.

Wherein, the transmission of incremental data can send routine by kernel to realize；Kernel reads routine for read increasing Amount data submit to kernel and send routine, and kernel sends routine and received incremental data is sent to backup server.

On rule opportunity based on the write IO request being locally trapped, continuous data protection, which can be divided into, to be synchronized and asynchronous two kinds Mode.Kernel reads routine in the present embodiment and kernel sends routine and can carry out incremental data reading and hair using asynchronous mode It send.

Wherein, the basic step of synchronous mode are as follows: 1, upper layer submission write IO request.2, Kernel Filtering driver intercepting and capturing are write I/O request.3, kernel sends routine and sends backup server for monitoring record.4, Kernel Filtering driver is by write IO request It is submitted to the write-in that the driving of disk lower layer carries out data.The advantages of synchronous mode is without local cache.But IO is write due to each It requests corresponding data to be first sent to backup server and is just submitted to local disk, under the IO performance meeting of local disk is serious Drop, the time-consuming expense of each write IO request increase 2 times or more for no filtration drive is interfered.In addition synchronous mode height The network bandwidth between backup target machine and backup server is relied on, therefore the mode can not usually be applied to long-distance disaster, only Suitable for local backup, with scenes such as city backup, High Availabitities；If being used for long-distance disaster, it is necessary to set up private network.

As shown in figure 3, the basic step of asynchronous mode are as follows: write IO request is submitted on S1, upper layer.S2, Kernel Filtering drive journey Sequence intercept and capture write IO request, and construct monitoring record (monitoring record include { offset, length, timestamp }+data, it may be assumed that Disk offset, data length, timestamp and data).Monitoring record is submitted to local cache by S3, Kernel Filtering driver Disk.Write IO request is submitted to the driving of disk lower layer by S4, Kernel Filtering driver.S5, kernel read routine from local cache Incremental data is extracted in disk and is sent to kernel sends routine.S6, kernel send routine and incremental data are sent to backup services Device.The advantages of asynchronous mode is: relatively small on the influence of the IO performance of monitoring panel, each write IO request time-consuming expense increases by 1 times Left and right is fewer than the time-consuming expense of synchronous mode；And asynchronous mode by local cache disk come smooth network bandwidth and disk speed Difference, therefore can be used for remote data backup, it can largely tolerant network postpone, can be applied to strange land appearance Calamity.But since asynchronous mode needs to store data, it is therefore desirable to additional local cache disk, if backup target There is no additional disc cache on machine, can not just implement continuous data protection scheme or needs by the external storage such as FC, ISCSI To provide disc cache.In addition, if disc cache capacity is too small or network bandwidth between backup target machine and backup server is remote Lower than the average data writing speed of monitoring panel, disc cache is easy to that cache overflow occurs.

In view of the above-mentioned problems, the continuous data protection method of no local cache provided in this embodiment, incremental metadata institute The space occupied is small, and the memory of target backup machine stores enough.Continuous data protection method phase with tradition based on local cache Than, without additional disk as disc cache, do not need the external storages such as FC, ISCSI yet, can to backup target machine and Complete machine implements continuity data protection.In addition, the present invention can tolerate low bandwidth compared with the continuous data protection of synchronous mode Network environment supports strange land data disaster tolerance.

In one embodiment, it after described the step of obtaining incremental metadata according to the data backup requests, also wraps It includes: in the increment record queue that the incremental metadata is added in memory；It is described according to the incremental metadata from The step of corresponding incremental data is read in disk, comprising: extract the incremental metadata from the increment record queue；Root Corresponding incremental data is read from disk according to extracted incremental metadata.

Wherein, increment record queue refers to for recording increment metadata and guarantees that kernel reads routine and reads in order Incremental metadata in increment record queue.And kernel reading routine incrementally metadata can be added to increment record queue Sequencing carry out the reading of incremental metadata.

Due to memory resource limitation, if the corresponding data of caching I/O request, memory are easy to overflow in memory.Therefore Traditional CDP scheme needs additional disk as disc cache for caching monitoring data.

In the present embodiment, incremental metadata is stored in increment record queue, and to be located at destination disk standby for increment record queue In the memory of part machine.The occupied space of incremental metadata is smaller, and (metadata information of 32 bytes can indicate 4K~512K Data, space hold is only the 1/16384~1/128 of traditional scheme), therefore can be in the sufficient amount of increment of memory cache Metadata.Therefore continuous data protection method provided in an embodiment of the present invention can realize continuous data protection without local cache, The convenience of continuous data protection can be effectively improved.

In one embodiment, the incremental metadata is deviated according to the disk in the data backup requests and data are long Degree obtains；Before the step incremental metadata being added in the increment record queue in memory, further includes: Incremental metadata is written on the corresponding sector position of variation sector bitmap；The corresponding sector position is deviated according to the disk It is determined with the data length；When detect the variation sector bitmap there are overlapping sector position when, will with it is described it is Chong Die fan The incremental metadata of the corresponding new write-in in position is added in conflict record queue；Wherein, the overlapping sector position is variation fan The incremental metadata being newly written in the bitmap area sector position be overlapped with history incremental metadata.

Change sector bitmap and passes through bitmap (bitmap) one monitoring panel of complete map, each (bit) of bitmap Indicate a sector of monitoring panel.Under original state, each of bitmap is all 0；When sector bitmap is set, pass through increment member number The start bit and stop bits that bitmap can be positioned according to the offset of corresponding disk and length, then set 1 for this section.? Can also check whether section has position before being set as 1 has been 1, if all positions are all 1, indicates that generation is completely overlapped, such as Fruit part position is 1, and expression partly overlaps.The completely overlapped repetition sector that is mainly used for disappears again, and the embodiment of the present invention mainly relates to And be the case where partly overlapping.These sectors being overlapped are exactly above-mentioned overlapping sector, refer to repeatedly being written same Sector.

In case where in Fig. 4, to illustrate the adding procedure of the incremental metadata in conflict record queue: increasing Measure (void with the arrow in such as Fig. 4 on the S1 of metadata record (i.e. incremental metadata, such as 401 in Fig. 4) R2 write-in sector 402 Line), sector S1 is also written in increment record (i.e. incremental metadata) R5 later；Increment records R4 and sector S6 is written, and increment is remembered later Sector S6 is also written in record R6；Increment records R8 and sector S10 is written, and sector S10 is also written in increment record R9 later.That is, The sector zone of the increment record write-in generated afterwards includes that more early increment records the sector having been written into, i.e. generation sector overlapping (is deposited In overlapping sector position), there are IO to conflict with other increments record by increment record R5, R6, R9 at this time, wherein increment record R5 and increasing Amount record R2 conflict, increment record R6 and increment record R4 conflict, and increment records R9 and increment record R8 conflict.It is recorded with increment The increment record R4 and R8 that R6 and R9 generates conflict is not yet read, and increment record R6 and R9 is conflict situation at this time, corresponding to increase Amount metadata record is added in conflict record queue 403.

Therewith whether Kernel Filtering driver can check constructed incremental metadata after constructing incremental metadata There are sector overlappings for the preceding record of incremental metadata earlier, and incremental metadata record is then added to increment record queue.Such as There is overlapping in fruit, which records while being added to conflict record queue, indicate entry into conflict situation.

It is the reading and transmission for using asynchronous mode to carry out incremental data since kernel reads routine and kernel transmission routine, Same position may occur multiple IO write-in and be read before being synchronized to backup server on disk, and then increment is caused to be remembered It records corresponding incremental data and is in inconsistent state.The present embodiment introduces variation sector bitmap to store incremental metadata, energy table The place whether data of some address in monitoring panel are repeatedly written and are read, and clashed is shown and passes through overlapping fan Position indicates.Kernel, which reads routine, can know that corresponding address is repeatedly written by overlapping sector position, and be written afterwards The data cover that data will may first be written, therefore data can not be read from corresponding address or by read data Labeled as colliding data or wrong data, the accuracy of read incremental data can be effectively ensured.In addition, such operation side Formula can detect that increment records the inconsistent state of corresponding incremental data, which can influence the reading of data in magnetic disk It takes and possible data is restored.

In one embodiment, described the step of reading corresponding incremental data from disk according to incremental metadata it Afterwards, further includes: remove corresponding incremental metadata in the overlapping sector position and conflict record queue；Traversal conflict record Conflict incremental metadata in queue, and judge to whether there is and the incremental metadata pair that conflicts in the bitmap of the variation sector The overlapping sector position answered；When it be not present, determine that the conflict incremental metadata is read in advance；By the conflict increment member number According to being added in inconsistency record queue；Wherein, the inconsistency record queue be used for from disk read incremental data it Preceding carry out consistency check；The consistency check includes determining the increment for needing the incremental data read and being written in disk The consistency of data.

Incremental metadata does not represent incremental data in the presence of conflict and is in inconsistent state, only when the sector of sending overlap In incremental metadata when being read in advance, there are the incremental metadatas of conflict just can switch to inconsistent state by conflict situation. For according to the example in a upper embodiment, as shown in figure 4, increment record R5 and increment record R2 conflict, when kernel is read When routine records R2 reading sector S1 according to increment, the data that sector S1 includes are really that increment record R5 is written, i.e. increment The data of record R5 write-in are read in advance, and the corresponding data of increment record R3 should record the number of R5 prior to increment According to being synchronized to Backup Data, therefore increment record R5 switchs to inconsistent state by conflict situation.Corresponding incremental metadata needs It is to be added to arrive inconsistency record queue 404.It is when inconsistent queue is empty, that is, standby when inconsistent incremental metadata is not present Part data are in disk level consistent state.

In addition, the incremental metadata that can be in write-in variation sector bitmap is also possible to increasing corresponding with incremental metadata Measure data.Further, when sector S100 is written for the identical disk sector of two increment record write-ins, such as increment record R1 Data AAAA, before increasing record R1 and being read, increment records R20 and data BBBB is written in sector S100.Increment is remembered at this time R1 and increment record R20 conflict are recorded, since increment record R1 is first written, reading routine also can first handle increment record R1.Work as reading When routine being taken to read data according to R1, what is actually read is the data of R20 write-in, therefore the data of R20 are read in advance, because This R20 is transformed into inconsistency record queue, and until R20 is read, data are restored to consistent state again.

Wherein, after completing data and continuously protecting, when if necessary to carry out data recovery；By the incremental number backed up According to being not corresponding with incremental metadata, therefore it is registered as inconsistent Backup Data and can be not used in data recovery.And it is right It in inconsistent Backup Data, can reacquire in other manners, for example, the data in inconsistency record queue are by independence Ground is stored on some address section of backup server, by read the data on this address section can be achieved with it is inconsistent Backup Data recovery.Such processing mode can guarantee the accuracy of restored data.

The present embodiment records the data read in advance by setting inconsistency record queue, can read in kernel Routine carries out judging whether the data in disk are read in advance when reading data, i.e., stores in the data and disk of required reading Data it is whether consistent.The read-write state of data in disk can be got in real time, and when incremental data is read in advance, is passed through Incremental metadata in inconsistency record queue controls the recoveries of data, can guarantee the accuracy of restored data.

In one embodiment, the data backup requests include write IO request；It is described to add the incremental metadata To before the step in the increment record queue being located in memory, further includes: deviated according to the disk in the write IO request true Corresponding incremental data is written in said write address by the writing address for determining disk.

Wherein, the write-in for carrying out incremental data can be driven by disk lower layer to realize.Lower layer's disk drive is according to writing The corresponding disk offset of I/O request is addressed, magnetic disk media (monitoring panel) is then write data into as unit of sector, i.e., in fact The rule of existing data.It should be noted that in embodiments of the present invention, it also includes that will correspond to that write IO request, which is written in disk, Incremental data be written to the step in disk.

The present embodiment can save in the disk of write IO request write-in write IO request, while guaranteeing that data are accurate Ground is written on accurate Disk Locality.

In one embodiment, described that corresponding incremental data is read from disk according to the incremental metadata；By institute The incremental data of reading is sent to backup server, to trigger the step that the backup server backs up the incremental data Suddenly, comprising: extract the first incremental metadata from the increment record queue, judge to correspond to according to first incremental metadata Incremental data whether reach single send the upper limit；If it is not, first incremental metadata is merged with the second incremental metadata, Obtain third incremental metadata；Wherein, second incremental metadata in increment record queue with the first increment member number According to adjacent；Corresponding incremental data is read from disk according to the third incremental metadata；Read incremental data is added Be added in transmit queue, with trigger the transmit queue by the read incremental data be sent to backup server carry out it is standby Part.

Wherein, the process that the first incremental metadata merges with the second incremental metadata can be realized by following steps: Check increment record queue in the first incremental metadata next incremental metadata (the second incremental metadata) whether and currently Reading position is adjacent backward；If it is, the first incremental metadata and the second incremental metadata are merged.Here merging Can refer to will place before and after two incremental metadatas, when carrying out the reading of incremental data, respectively according to the first incremental metadata Corresponding first incremental data and the second incremental data are obtained with the second incremental metadata.Later, by the first incremental data and Two incremental datas are all added in transmit queue.

Certainly, the number of the second incremental metadata can be multiple.That is, the length in incremental data is not up to single transmission In the case where the upper limit, multiple adjacent incremental metadatas backward can be merged into an incremental metadata and be read out and send out It send.

The present embodiment merges adjacent incremental metadata, in turn in the case where being no more than the single transmission upper limit The backup for carrying out data, can effectively improve the efficiency of data backup, the continuous protection of data is better achieved.

In one embodiment, the step for reading corresponding incremental data from disk according to the incremental metadata Suddenly, comprising: in current reading cycle, when the number of the incremental metadata is more than given threshold, stopping is read from disk Incremental data is taken, until next read cycle reaches.

When carrying out data write-in, Kernel Filtering driving can construct an incremental metadata and is cached in memory Incremental data, is then written on monitoring panel by increment record queue.Kernel reads routine according to incremental metadata from monitoring panel Upper reading incremental data submits to kernel transmission routine and is sent to backup server.If IO frequently (frequently receives IO to ask Ask), kernel, which reads routine, can compete disk resource with other systems application, and then the IO delay for causing other systems to be applied Increase, thus this programme introduce write performance override mechanism: when IO is frequent, reduce kernel read routine working frequency (such as: Kernel reads routine and this read action is postponed till next read cycle processing), it is ensured that other systems application is preferential to be obtained Disk resource completes I/O operation.Wherein, the frequent method of IO is write in judgement are as follows: judges that write IO request is corresponding in each read cycle Whether incremental metadata number is more than given threshold；If it exceeds given threshold, is judged to writing IO frequent.

In addition, incremental metadata can be continuously increased, and then cause to increase if kernel is read routine and worked with low frequency always Record queue is measured to overflow.Therefore the present embodiment limits the length of an incremental metadata record queue on the basis of write performance is preferential Threshold value is spent, when reaching this threshold value, kernel is read routine and worked with normal frequency, it is ensured that incremental metadata record is located as early as possible Reason, avoids increment record queue from occupying excessive memory.For example, the overhead thresholds of incremental metadata in memory can be limited. Assuming that this overhead thresholds is 50M, buffer data size is 4GB~512GB.In this case, when in increment record queue It deposits expense and reaches this overhead thresholds, even if IO is frequent, still read incremental data.

In one embodiment, as shown in figure 5, providing a kind of read method of incremental data, comprising the following steps:

S501, incremental metadata is obtained；The incremental metadata is for describing incremental data.

S502, the write state label for obtaining the incremental metadata；The write state label is for characterizing corresponding increment Whether data are successfully written in disk.

The case where being successfully written to disk for incremental data, write state label can be to write complete label；For increment Data are not successfully written to the case where disk (for example, the case where having not been written to or be written failure), and write state label can be Label to be done.

Specifically, write state label includes: to write complete label and label to be done；According to extracted incremental metadata with And write state label, the step of corresponding incremental data is read from disk, comprising: when the incremental metadata is corresponding Write state reads corresponding incremental data according to extracted incremental metadata labeled as when writing complete label from disk；When When the corresponding write state of the incremental metadata is labeled as label to be done, judge whether next read cycle reaches；If so, Corresponding incremental data is read from disk according to extracted incremental metadata.

In addition, can not also be carried out again when next read cycle reaches when write state is labeled as label to be done The reading of data, such as: it can be marked with real-time detection write state, become writing when detecting that write state is marked from label to be done When at label, with regard to carrying out the reading of incremental data.

Kernel reads routine and reads data from disk according to incremental metadata, therefore incremental metadata is being added to increasing It before amount record queue needs that the write IO request being trapped first is submitted to lower layer's disk drive, otherwise reading routine may read Get old data in magnetic disk.That is, can also include: to receive before the step of obtaining the write state label of the incremental metadata Write IO request；It include disk offset in the write IO request；It is deviated according to the disk in the write IO request and determines disk Corresponding incremental data is written in said write address by writing address.

S503, when the write state is labeled as label is write complete, read from disk according to the incremental metadata pair The incremental data answered.

The kernel of certain versions may merge I/O request in filter drive program lower layer, so even ensuring first IO to be submitted to ask Lower layer's disk drive is sought, then data is read further according to incremental metadata and still is possible to read the read I O request of routine submission It is responded prior to the write IO request previously submitted by lower layer's disk drive.Each incremental metadata record, which needs to increase, thus writes shape State label, write state label are updated after the corresponding data rule of write IO request.Routine is read to be remembered according to incremental metadata Record needs first to check that this marks whether to be updated to completion status when reading data, if do not completed, is deferred to next reading week Phase processing.Target backup machine can be made to read correct incremental data from disk, guarantee continuous data protection it is normal into Row.

In one embodiment, as shown in fig. 6, providing a kind of read method of data, comprising the following steps:

S601, incremental metadata is obtained；The incremental metadata is for describing incremental data.

S602, the consistency label for obtaining target data；Wherein, the target data is incremental metadata description Incremental data, the consistency of incremental data of the consistency label for describing to be written in the target data and disk.

Wherein, target data can refer to the incremental data for needing to read in disk.Consistency label can pass through aforementioned reality The information in the inconsistency record queue in example is applied to determine.

After the step of obtaining incremental metadata, further includes: the incremental metadata is added to the increasing in memory It measures in record queue；It is described to work as according to the incremental data one being written in the determining target data of consistency label and disk When cause, the step of corresponding incremental data is read from disk, comprising: when determining the number of targets according to consistency label According to it is consistent with the incremental data being written in disk when, extract the incremental metadata from the increment record queue；According to institute The incremental metadata of extraction reads corresponding incremental data from disk.

S603, when determining that the incremental data being written in the target data and disk is consistent according to consistency label When, corresponding incremental data is read from disk.

The realization process of the step may include: to determine the consistency label when inconsistency record queue is empty It is consistent；Corresponding incremental data is read from disk according to extracted incremental metadata；When inconsistency record queue is not When empty, determine the consistency labeled as inconsistent；The inconsistency record queue is emptied, is by the consistency flag update Unanimously；And corresponding incremental data is read from disk according to extracted incremental metadata.

Wherein, the purpose for emptying inconsistency record queue is that target backup machine is prevented to be in inconsistent state for a long time, is given The continuous data protection of target backup machine impacts.After emptying inconsistency record queue, it was originally recorded in inconsistency record team Incremental metadata in column can be directly deleted, and can also be sent routine by kernel and be sent to backup server, with storage In backup server in some independent memory space.

The present embodiment marks the incremental data for determining and needing the incremental data read and being written in disk according to consistency It is whether consistent, and then corresponding incremental data is read from disk, target backup machine can be made to read from disk correctly Incremental data.

In one embodiment, before the step of obtaining the consistency label of target data, further includes: IO is write in reception to be asked It asks；It include disk offset in the write IO request；When the consistency is labeled as consistent, according in the write IO request Disk deviates the writing address for determining disk, and corresponding incremental data is written in said write address；When the consistency Labeled as it is inconsistent when, judge whether to receive sleep END instruction；When being, deviated according to the disk in the write IO request Corresponding incremental data is written in said write address by the writing address for determining disk.The present embodiment is only in consistency Labeled as it is consistent when, just carry out write IO request write-in.

When filter drive program in the present embodiment intercepts write IO request every time, need to check that consistency marks, if Into consistent sexual cycle is forced, then write operation is delayed by sleep movement.The purpose for delaying write operation is to reduce increment member The formation speed of data, while the reading and transmission of high-speed are kept, make Backup Data enter consistent state as early as possible with this.

Certainly, in other embodiments, before carrying out write IO request write-in and reading data, if it is determined that consistency Labeled as inconsistent, in addition to waiting next write-in or the arrival of read cycle, the shape of consistency label also can be monitored in real time State just carries out corresponding data write-in and read operation when consistency label becomes consistent.

In one embodiment, when increment record collision probability is higher, Backup Data may be for a long time in inconsistent State.Increase thus and forces consistency timeout mechanism.That is, when inconsistency record queue is empty, it is described according to extracted increasing Metadata is measured before the step of reading corresponding incremental data in disk, further includes: when according to current time update consistency Between.Further, when inconsistency record queue is not sky, described the step of emptying the inconsistency record queue, comprising: really The time difference for determining current time Yu consistency time empties described when the time difference is greater than preset time threshold Inconsistency record queue.It reads when routine reads data and checks data consistency, if data are in consistent point, update consistency Time is current time；If be in non-uniform point, check current time unanimously put away from last time whether time-out, if time-out if set It sets and forces consistency label.The present embodiment can prevent Backup Data to be in inconsistent shape for a long time by consistency timeout mechanism State.

In one embodiment, as shown in fig. 7, providing a kind of continuous data protection method of no local cache, including Following steps:

S701, multiple data backup requests are obtained, is obtained according to the multiple data backup requests corresponding multiple initial Incremental metadata；Wherein, the initial gain amount metadata is for describing incremental data.

When disk is written in bulk continuous data, kernel may be split as multiple write IO requests, these are write IO and is asked Efficiency of backing up in realtime can be greatly improved by asking corresponding incremental metadata to merge into a reading and sending action.Therefore, it is connecing When receiving multiple write IO requests, corresponding initial gain amount metadata can be converted into according to these write IO requests, and then corresponding Incremental metadata.

S702, the multiple initial gain amount metadata is merged, obtains incremental metadata.

Initial gain amount metadata can be deviated according to the disk in the data backup requests and data length obtains.

This step can be realized by following steps: according to disk offset and the data length, to described more A initial gain amount metadata merges.

Further, the step of the multiple initial gain amount metadata being merged, obtains incremental metadata, comprising: The multiple initial gain amount metadata is merged, obtains merging initial gain amount metadata；Obtain the merging initial gain amount The merging length of metadata；When the merging length is greater than preset merging length threshold, terminate to the initial gain amount member The merging initial gain amount metadata is determined as the incremental metadata by the merging of data.Wherein, merging length threshold can be with It is arranged according to the actual situation, for example, can be 1024 sectors.

The specific implementation of this step can be with are as follows: to avoid single from reading and sending that data volume is excessive, and setting merges length Degree threshold value is 1024 sectors.As shown in figure 8, increment record continuous situation include 6 kinds of scenes, wherein (1), (3), (4) be to Continuous afterwards, (2), (5), (6) are continuous forward.The present embodiment only considers situation continuous backward, i.e. Offset_A< < Offset_B, and Offset_A+Length_A> > Offset_B, and Offset_B+Length_B-Offset_A> > T_Length, wherein Offset_AFor the disk offset of current delta record, Offset_BFor the disk offset of next increment record, Length_ATo work as The data length of preceding increment record, Length_BFor the length of next increment record, T_LengthTo merge length threshold.It will backward Continuous initial gain amount metadata merges into an incremental metadata.

S703, corresponding incremental data is read from disk according to the incremental metadata.

Realization about the step can refer to the correspondence step S202 of above-mentioned implementation, not repeat.

S704, read incremental data is sent to backup server, to trigger the backup server to the increasing Amount data are backed up.

Realization about the step can refer to the correspondence step S203 of above-mentioned implementation, not repeat.

The corresponding initial gain amount metadata of multiple data backup requests is merged, obtains increment member number by the present embodiment According to the incremental metadata obtained according to merging reads corresponding incremental data and is sent to backup server and backed up, and can have Effect improves the efficiency backed up in realtime.

The above method in order to better understand, as shown in figure 9, company of the present invention detailed below without local cache The application example of continuous data guard method.

Write IO request is submitted on S901, upper layer.

S902, filter drive program intercept and capture write IO request, extract disk offset therein, data length, IO timestamp Etc. information, and construct incremental metadata record (offset, length, timestamp).

Write IO request is submitted to the driving of disk lower layer by S903, filter drive program, and disk lower layer drives write IO request It is written in monitoring panel.

S904, target backup machine polishing incremental metadata record so that incremental metadata record the offset of corresponding disk and Data length meets 4KB alignment.And corresponding disk offset and data length setting variation sector are recorded according to incremental metadata Bitmap checks for overlapping sector position in setting up procedure.

S905, incremental metadata record is added in the increment record queue in memory.

If detecting overlapping sector position in S906, S904, corresponding incremental metadata record is added to conflict note It records in queue.

S907, kernel read routine and extract an incremental metadata record A from increment record queue.

S908, the incremental metadata record A in overlapping sector position and conflict record queue is removed；When the conflict of judgement is remembered When incremental metadata record A in record queue is read in advance, incremental metadata record A is added to inconsistency record queue In.

Whether not up to S909, kernel read incremental metadata acquired in routine inspection and record corresponding incremental data list The secondary transmission upper limit.If not up to single sends the upper limit, obtains an adjacent incremental metadata and record and carry out incremental metadata The merging of record.After through data consistent check, increment is read from disk according to the incremental metadata record after merging Data.If reaching single sends the upper limit, read from disk after through data consistent check according to incremental metadata record Incremental data.

S910, kernel reading routine will be added in transmit queue from the incremental data read in disk.

S911, kernel send routine and extract incremental data recording and sending to backup server from transmit queue.

The present embodiment, the occupied space of incremental metadata is small, and the memory of target backup machine stores incremental metadata enough, Therefore additional local cache is not needed, the protection of continuous data can be made more convenient.Meanwhile by variation sector bitmap, Conflict record queue carries out that collision detection is consistent with data to put inspection with inconsistency record queue, can guarantee continuous data protection Orderly progress.

The read method and number of the continuous data protection method without local cache, incremental data in the embodiment of the present invention According to read method in be related to the step of can be optionally combined.The specific embodiment of a certain step is applied equally to Other include the embodiment of same steps.

It should be noted that for the various method embodiments described above, describing for simplicity, it is all expressed as a series of Combination of actions, but those skilled in the art should understand that, the present invention is not limited by the sequence of acts described, because according to According to the present invention, certain steps can use other sequences or carry out simultaneously.

Based on thought identical with the continuous data protection method without local cache in above-described embodiment, the present invention is also mentioned For the continuous data protection system of no local cache, which can be used for executing the continuous data protection side of above-mentioned no local cache Method.For ease of description, in the structural schematic diagram of the continuous data protection system embodiment of no local cache, illustrate only with The relevant part of the embodiment of the present invention, it will be understood by those skilled in the art that the restriction of schematic structure not structure paired systems, it can To include perhaps combining certain components or different component layouts than illustrating more or fewer components.

In one embodiment, as described in Figure 10, the continuous data protection system of no local cache includes filtration drive mould Block 1001, read module 1002 and sending module 1003, detailed description are as follows: filtration drive module 1001, for obtaining data Backup request；Incremental metadata is obtained according to the data backup requests；Wherein, the incremental metadata is for describing incremental number According to.Read module 1002, for reading corresponding incremental data from disk according to the incremental metadata.And send mould Block 1003, for read incremental data to be sent to backup server, to trigger the backup server to the increment Data are backed up.The present embodiment, the occupied space of incremental metadata is small, and the memory of target backup machine stores increment member enough Data, therefore additional local cache is not needed, the protection of continuous data can be made more convenient.

In one embodiment, further includes: the first record queue adding module, for the incremental metadata to be added to In increment record queue in memory；Read module 1002, comprising: the first meta-data extraction submodule is used for from described The incremental metadata is extracted in increment record queue；First reading data submodule, for according to extracted increment member number According to reading corresponding incremental data from disk.

In one embodiment, the incremental metadata is deviated according to the disk in the data backup requests and data are long Degree obtains；Further include: sector bitmap writing module, for the corresponding sector position of variation sector bitmap to be written in incremental metadata On；The corresponding sector position is determined according to disk offset and the data length；Second record queue adds mould Block, for when detecting that the variation sector bitmap there are when the position of overlapping sector, corresponding with overlapping sector position will newly write The incremental metadata entered is added in conflict record queue；Wherein, the overlapping sector position is newly written to change in the bitmap of sector Incremental metadata it is be overlapped with history incremental metadata sector position.

In one embodiment, further includes: data dump module, for the overlapping sector position and conflict to be recorded team Corresponding incremental metadata is removed in column；It is overlapped judgment module, for traversing the conflict incremental metadata in conflict record queue, And judge in the bitmap of the variation sector with the presence or absence of overlapping sector position corresponding with the conflict incremental metadata；Inconsistent team Column adding module, for when it be not present, determining that the conflict incremental metadata is read in advance；By the conflict increment member number According to being added in inconsistency record queue；Wherein, the inconsistency record queue be used for from disk read incremental data it Preceding carry out consistency check；The consistency check includes determining the increment for needing the incremental data read and being written in disk The consistency of data.

In one embodiment, the data backup requests include write IO request；Further include: the first Data write. module, For deviating the writing address for determining disk according to the disk in the write IO request, corresponding incremental data is written to described In writing address.

In one embodiment, further includes: upper limit judgment module is sent, for extracting the from the increment record queue One incremental metadata judges whether corresponding incremental data reaches single and send the upper limit according to first incremental metadata；Member Data combiners block is used to that it is first to obtain third increment if it is not, first incremental metadata is merged with the second incremental metadata Data；Wherein, second incremental metadata is adjacent with first incremental metadata in increment record queue；First increment Data read module, for reading corresponding incremental data from disk according to the third incremental metadata；Data send mould Block, for read incremental data to be added in transmit queue, to trigger the transmit queue for the read increasing Amount data are sent to backup server and are backed up.

In one embodiment, read module 1002 are also used in current reading cycle, when the incremental metadata When number is more than given threshold, incremental data is read in stopping from disk, until next read cycle reaches.

In one embodiment, as described in Figure 11, the reading system of incremental data includes filtration drive module 1101, label Obtain module 1102 and read module 1103.Detailed description are as follows: filtration drive module 1101, for obtaining incremental metadata； Wherein, the incremental metadata is for describing incremental data；Label obtains module 1102, for obtaining the incremental metadata Write state label；The write state label is for characterizing whether corresponding incremental metadata is successfully written in disk；And it reads Modulus block 1103, for being read from disk according to the incremental metadata when the write state is labeled as label is write complete Corresponding incremental data.The present embodiment is marked by setting write state, can characterize whether corresponding incremental data is successfully written to In disk, therefore target backup machine can be made to read correct incremental data from disk.

In one embodiment, further includes: I/O request receiving module, for receiving write IO request；In the write IO request It include disk offset；Second Data write. module determines writing for disk for deviating according to the disk in the write IO request Enter address, corresponding incremental data is written in said write address.

In one embodiment, further includes: third record queue adding module, for the incremental metadata to be added to In increment record queue in memory；Read module 1103, comprising: the second meta-data extraction submodule is used for from described The incremental metadata is extracted in increment record queue；Second reading data submodule, for according to extracted increment member number Accordingly and the write state marks, and corresponding incremental data is read from disk.

In one embodiment, the write state label includes: to write complete label and label to be done；Read module 1103, comprising: the second incremental data read module, for being labeled as writing complete mark when the corresponding write state of the incremental metadata It clocks, corresponding incremental data is read from disk according to extracted incremental metadata；Third incremental data read module is used In when the corresponding write state of the incremental metadata is labeled as label to be done, judge whether next read cycle reaches； If so, reading corresponding incremental data from disk according to extracted incremental metadata.

In one embodiment, as described in Figure 12, the reading system of data includes filtration drive module 1201, label acquisition Module 1202 and read module 1203.Detailed description are as follows: filtration drive module 1201, for obtaining incremental metadata；It is described Incremental metadata is for describing incremental data；Label obtains module 1202, and the consistency for obtaining target data marks；Its In, the target data is the incremental data of incremental metadata description, and the consistency label is for describing the target The consistency for the incremental data being written in data and disk；And read module 1203, for being marked when according to the consistency When determining that the incremental data being written in the target data and disk is consistent, corresponding incremental data is read from disk.This reality Example is applied, whether the incremental data being written in the incremental data and disk for marking determining needs to read according to consistency is consistent, into And corresponding incremental data is read from disk, target backup machine can be made to read correct incremental data from disk.

In one embodiment, further includes: the 4th record queue adding module, for the incremental metadata to be added to In increment record queue in memory；Read module 1203, comprising: third meta-data extraction submodule, for working as basis When the consistency label determines that the incremental data being written in the target data and disk is consistent, from the increment record queue It is middle to extract the incremental metadata；Third reading data submodule, for being read from disk according to extracted incremental metadata Take corresponding incremental data.

In one embodiment, read module 1203, comprising: the 4th incremental data read module, for working as inconsistent note When record queue is empty, determine the consistency labeled as consistent；Correspondence is read from disk according to extracted incremental metadata Incremental data；5th incremental data read module, for determining the consistency mark when inconsistency record queue is not sky It is denoted as inconsistent；The inconsistency record queue is emptied, is consistent by the consistency flag update；And according to extracted increasing Amount metadata reads corresponding incremental data from disk.

In one embodiment, when inconsistency record queue is empty, further includes: consistency time update module is used for According to the current time update consistency time.

In one embodiment, when inconsistency record queue is not sky, the 5th incremental data read module is also used to really The time difference for determining current time Yu consistency time empties described when the time difference is greater than preset time threshold Inconsistency record queue.

In one embodiment, as described in Figure 13, the continuous data protection system of no local cache includes filtration drive mould Block 1301, merging module 1302, read module 1303 and filtration drive module 1301, for obtaining multiple data backup requests, Corresponding multiple initial gain amount metadata are obtained according to the multiple data backup requests；Wherein, the initial gain amount metadata For describing incremental data；Merging module 1302 obtains increment member for merging to the multiple initial gain amount metadata Data；Read module 1303, for reading corresponding incremental data from disk according to the incremental metadata；And it sends Module 1304, for read incremental data to be sent to backup server, to trigger the backup server to the increasing Amount data are backed up.The corresponding initial gain amount metadata of multiple data backup requests is merged, is increased by the present embodiment Measure metadata, according to merge obtained incremental metadata read corresponding incremental data and be sent to backup server carry out it is standby Part, the efficiency backed up in realtime can be effectively improved.

In one embodiment, the initial gain amount metadata deviates sum number according to the disk in the data backup requests It is obtained according to length；Merging module 1302 is also used to according to disk offset and the data length, to the multiple initial increasing Amount metadata merges.

In one embodiment, merging module 1302, comprising: initial metadata merges submodule, for the multiple Initial gain amount metadata merges, and obtains merging initial gain amount metadata；Merge length acquisition submodule, it is described for obtaining Merge the merging length of initial gain amount metadata；Metadata determining module, for being greater than preset merging when the merging length When length threshold, terminate the merging to the initial gain amount metadata, the merging initial gain amount metadata is determined as described Incremental metadata.

It should be noted that the continuous data protection system of no local cache of the invention and no local cache of the invention Continuous data protection method correspond, above-mentioned no local cache continuous data protection method embodiment illustrate skill Suitable for the embodiment of the continuous data protection system of no local cache, particular content can join for art feature and its advantages See the narration in embodiment of the present invention method, details are not described herein again, hereby give notice that.

In addition, in the embodiment of the continuous data protection system without local cache of above-mentioned example, each program module Logical partitioning is merely illustrative of, and can according to need in practical application, such as the configuration requirement of corresponding hardware or soft The convenient of the realization of part considers, above-mentioned function distribution is completed by different program modules, i.e., by the company of the no local cache The internal structure of continuous data protection system is divided into different program modules, to complete all or part of function described above Energy.

In one embodiment, a kind of computer equipment is provided, which can be server, internal junction Composition can be as shown in figure 14.The computer equipment include by system bus connect processor, memory, network interface and Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating The database of machine equipment is for storing incremental metadata, incremental data etc..The network interface of the computer equipment is used for and outside Terminal by network connection communication.A kind of consecutive numbers of no local cache is realized when the computer program is executed by processor According to guard method.

It will be understood by those skilled in the art that structure shown in Figure 14, only part relevant to application scheme The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set Standby may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.

In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory And the computer program that can be run on a processor, processor performs the steps of acquisition data when executing computer program standby Part request；Incremental metadata is obtained according to the data backup requests；Wherein, the incremental metadata is for describing incremental number According to；Corresponding incremental data is read from disk according to the incremental metadata；Read incremental data is sent to backup Server is backed up with triggering the backup server to the incremental data.

In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory And the computer program that can be run on a processor, processor performs the steps of when executing computer program obtains increment member Data；The incremental metadata is for describing incremental data；Obtain the write state label of the incremental metadata；The write state Label is for characterizing whether corresponding incremental data is successfully written in disk；When the write state is labeled as writing complete label When, corresponding incremental data is read from disk according to the incremental metadata.

In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory And the computer program that can be run on a processor, processor performs the steps of when executing computer program obtains increment member Data；The incremental metadata is for describing incremental data；Obtain the consistency label of target data；Wherein, the number of targets According to the incremental data described for the incremental metadata, the consistency label is for describing to write in the target data and disk The consistency of the incremental data entered；When according to the incremental number being written in the determining target data of consistency label and disk According to it is consistent when, corresponding incremental data is read from disk.

In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory And the computer program that can be run on a processor, processor performs the steps of when executing computer program obtains multiple numbers According to backup request, corresponding multiple initial gain amount metadata are obtained according to the multiple data backup requests；Wherein, described initial Incremental metadata is for describing incremental data；The multiple initial gain amount metadata is merged, incremental metadata is obtained；Root Corresponding incremental data is read from disk according to the incremental metadata；Read incremental data is sent to backup services Device is backed up with triggering the backup server to the incremental data.

In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated Machine program performs the steps of acquisition data backup requests when being executed by processor；Increased according to the data backup requests Measure metadata；Wherein, the incremental metadata is for describing incremental data；It is read from disk according to the incremental metadata pair The incremental data answered；Read incremental data is sent to backup server, to trigger the backup server to the increasing Amount data are backed up.

In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated Machine program performs the steps of acquisition incremental metadata when being executed by processor；The incremental metadata is for describing incremental number According to；Obtain the write state label of the incremental metadata；Write state label for characterize corresponding incremental data whether by It is successfully written in disk；When the write state is labeled as label is write complete, read from disk according to the incremental metadata Corresponding incremental data.

In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated Machine program performs the steps of acquisition incremental metadata when being executed by processor；The incremental metadata is for describing incremental number According to；Obtain the consistency label of target data；Wherein, the target data is the incremental data of incremental metadata description, The consistency of incremental data of the consistency label for describing to be written in the target data and disk；When according to described one When cause property label determines that the incremental data being written in the target data and disk is consistent, corresponding incremental number is read from disk According to.

In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated Machine program performs the steps of when being executed by processor obtains multiple data backup requests, is asked according to the multiple data backup It asks to obtain corresponding multiple initial gain amount metadata；Wherein, the initial gain amount metadata is for describing incremental data；To described Multiple initial gain amount metadata merge, and obtain incremental metadata；Correspondence is read from disk according to the incremental metadata Incremental data；Read incremental data is sent to backup server, to trigger the backup server to the increment Data are backed up.

It will appreciated by the skilled person that realizing all or part of the process in above-described embodiment method, being can It is completed with instructing relevant hardware by computer program, the program can be stored in a computer-readable storage and be situated between In matter, sells or use as independent product.The more specific example (non-exhaustive list) of computer-readable medium includes Below: there is the electrical connection section (electronic device) of one or more wirings, portable computer diskette box (magnetic device), arbitrary access Memory (RAM), read-only memory (ROM), erasable edit read-only storage (EPROM or flash memory), optical fiber dress It sets and portable optic disk read-only storage (CDROM).In addition, computer-readable medium, which can even is that, to be printed on it The paper of described program or other suitable media, because can be for example by carrying out optical scanner to paper or other media, then It edited, interpreted or is handled when necessary with other suitable methods electronically to obtain described program, then by it Storage is in computer storage.

It should be appreciated that each section of the invention can be realized with hardware, software, firmware or their combination.Above-mentioned In embodiment, software that multiple steps or method can be executed in memory and by suitable instruction execution system with storage Or firmware is realized.It, and in another embodiment, can be under well known in the art for example, if realized with hardware Any one of column technology or their combination are realized: having a logic gates for realizing logic function to data-signal Discrete logic, with suitable combinational logic gate circuit specific integrated circuit, programmable gate array (PGA), scene Programmable gate array (FPGA) etc..

The term " includes " of the embodiment of the present invention and " having " and their any deformations, it is intended that cover non-exclusive Include.Such as contain series of steps or the process, method, system, product or equipment of (module) unit are not limited to The step of listing or unit, but optionally further comprising the step of not listing or unit, or optionally further comprising for these The intrinsic other step or units of process, method, product or equipment.

Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited In contradiction, all should be considered as described in this specification.

The embodiments described above only express several embodiments of the present invention, should not be understood as to the invention patent range Limitation.It should be pointed out that for those of ordinary skill in the art, without departing from the inventive concept of the premise, Various modifications and improvements can be made, and these are all within the scope of protection of the present invention.Therefore, the scope of protection of the patent of the present invention It should be determined by the appended claims.

Claims

1. a kind of read method of incremental data, which comprises the following steps:

Obtain incremental metadata；The incremental metadata is for describing incremental data；

Obtain the write state label of the incremental metadata；Write state label for characterize corresponding incremental data whether by It is successfully written in disk；

When the write state is labeled as label is write complete, corresponding incremental number is read from disk according to the incremental metadata According to.

2. the read method of incremental data according to claim 1, which is characterized in that described to obtain the incremental metadata Write state label the step of before, further includes:

Receive write IO request；It include disk offset in the write IO request；

The writing address for determining disk is deviated according to the disk in the write IO request, corresponding incremental data is written to described In writing address.

3. the read method of incremental data according to claim 1, which is characterized in that the step for obtaining incremental metadata After rapid, further includes:

The incremental metadata is added in the increment record queue in memory；

It is described when the write state labeled as write complete label when, corresponding increasing is read from disk according to the incremental metadata The step of measuring data, comprising:

The incremental metadata is extracted from the increment record queue；

It is marked according to extracted incremental metadata and the write state, corresponding incremental data is read from disk.

4. the read method of incremental data according to claim 3, which is characterized in that the write state label includes to complete At label；

It is described to be marked according to extracted incremental metadata and the write state, corresponding incremental data is read from disk Step, comprising:

When the corresponding write state of the incremental metadata is labeled as label is write complete, according to extracted incremental metadata from magnetic Corresponding incremental data is read in disk；

When the corresponding write state of the incremental metadata is labeled as label to be done, judge whether next read cycle arrives It reaches；If so, reading corresponding incremental data from disk according to extracted incremental metadata.

5. the read method of incremental data according to claim 3, which is characterized in that the incremental metadata includes following At least one of: disk offset and data length；

Before the step incremental metadata being added in the increment record queue in memory, further includes:

Incremental metadata is written on the corresponding sector position of variation sector bitmap；The corresponding sector position is according to the disk What offset and data length determined；

When detecting the variation sector bitmap there are when the position of overlapping sector, by new write-in corresponding with overlapping sector position Incremental metadata is added in conflict record queue；Wherein, the overlapping sector position is the increasing being newly written in variation sector bitmap Measure the metadata sector position be overlapped with history incremental metadata.

6. the read method of incremental data according to claim 5, which is characterized in that it is described read from disk it is corresponding After the step of incremental data, further includes:

Corresponding incremental metadata in the overlapping sector position and conflict record queue is removed；

Traversal conflict record queue in conflict incremental metadata, and judge in the bitmap of the variation sector whether there is with it is described The corresponding overlapping sector position of the incremental metadata that conflicts；

When it be not present, determine that the conflict incremental metadata is read in advance；The conflict incremental metadata is added to not In consistent record queue；Wherein, the inconsistency record queue is used to carry out unanimously before reading incremental data in disk Property inspection；The consistency check includes determining to need the consistent of the incremental data that reads and the incremental data being written in disk Property.

7. the read method of incremental data according to any one of claims 1 to 6, which is characterized in that described from disk The step of reading corresponding incremental data, comprising:

In current reading cycle, when the number of the incremental metadata is more than given threshold, stop reading increasing from disk Data are measured, until next read cycle reaches.

8. a kind of reading system of incremental data characterized by comprising

Filtration drive module, for obtaining incremental metadata；Wherein, the incremental metadata is for describing incremental data；

Label obtains module, and the write state for when receiving data read request, obtaining incremental metadata marks；It is described to write Status indication is for characterizing whether corresponding incremental metadata is successfully written in disk；

And read module, for when the write state is labeled as label is write complete, according to the incremental metadata from disk It is middle to read corresponding incremental data.

9. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor Calculation machine program, which is characterized in that the processor is realized described in any one of claim 1 to 7 when executing the computer program Method the step of.

10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program The step of claim 1 to 7 described in any item methods are realized when being executed by processor.