CN1071770A - Be used for time zero according to the method and apparatus of backing up - Google Patents
Be used for time zero according to the method and apparatus of backing up Download PDFInfo
- Publication number
- CN1071770A CN1071770A CN92110772A CN92110772A CN1071770A CN 1071770 A CN1071770 A CN 1071770A CN 92110772 A CN92110772 A CN 92110772A CN 92110772 A CN92110772 A CN 92110772A CN 1071770 A CN1071770 A CN 1071770A
- Authority
- CN
- China
- Prior art keywords
- cpu
- scope
- data set
- write
- data collection
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1446—Point-in-time backing up or restoration of persistent data
- G06F11/1458—Management of the backup or restore process
- G06F11/1466—Management of the backup or restore process to make the backup process non-disruptive
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F12/00—Accessing, addressing or allocating within memory systems or architectures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Techniques For Improving Reliability Of Storages (AREA)
Abstract
Backing up of the specific data collection of compatibility can be by just enough constituting the time that a logic arrives physical address index carrying out one of hang-up on the embodiment time point on the DASD storage subsystem, in a CPU, carry out and use execution concurrence with CPU and carry out, subsequently this data set of reserve on this storage subsystem physically on basis predetermined or at random.Renewal to the application start of the specific data collection that do not duplicate at first is cushioned storage, constitutes appended document with affected data set, to upgrade directly and write on the storage subsystem, and appended document will be write on the storer with the order of backing up of index control.
Description
The data set that the present invention relates in external storage keeps continuous utilizability to the computer system (CPU) of access.More specifically, it relates to and duplicates the CPU that causes with this and use the extremely short pause of carrying out back up record concomitantly in external storage.
Data disposal system must have preparation not only from such as noise spike, software fault, media defect and write the destruction of path error to the data of being stored and recover, and preparation must be arranged from such as recovering the incidents of overall importance such as cpu power fault.The technology of the continuous utilizability of modal assurance data is one or more duplicate of composition data collection and the place that places them in a safety.This " reserve " process appears in the scope of the storage system that increases function day by day.
Application program with a kind of batch of (stream) pattern or mutual (affairs) pattern both one of on CPU, carry out.In batch mode, common one next application program ground is not interruptedly carried out.Interactive mode then is that multiplicity with the application program of drives interrupts or affairs is a feature.
The reserve strategy is a scheduling strategy.They had with the space and the time of the explanation of the frequency of the scope of data set and generation measures.No matter representing gamut upgraded, full reserve whether all duplicates.The data set part that changed after the last reserve (no matter full reserve or increment reserve) is only duplicated in an increment reserve.Back up to be illustrated in and make this and duplicate or a compatible view of the moment data of fast elephant.
Backup frequency is high more, backs up the current of approaching more reflection data and duplicates.Consider that data volume is big, reserve is not a kind of footy attended operation.Like this, handle with respect to other, the opportunity cost of the reserve on a large-scale multiple processing, multiprogramming equipment can be high.
When a CPU in stream or batch mode system during the reserve data, each process, task or applications wait.This meaning supports the process of stream or batch mode operation to pause in the perdurability of duplicating.The title that rises of this incident of giving is called " reserve window ".Opposite with batch mode, with log on as the basis or transaction management use and to handle with interactive mode.It changes in fact and eliminates " reserve window " by upgrading an online data collection and login concomitantly for they.Yet the latter is that a kind of compatibility " fuzzy " is backed up mode.That is, it is not the resembling soon of state of a data set/database on a single time point.It would be better to say that a login is need be to the further incident file of handling of described database.
The people's such as Wang of " the using a kind of computer based data set replica method of increment reserve strategy " by name of submitting on July 25th, 1989 common pending application USSN:07/385,647(IBM list of references SA9-89-043), show the reserve in a batch mode system of the increment strategy that uses a kind of correction.A kind of increment strategy of correction has only duplicated new data or the Data Update since reserve last time.It should be noted that to use in duplicating and pause.
As mentioned above, for the point early of setting up a compatibility in the system that logs on as the basis at one, be necessary by reset on data set of being concerned about or the database from once the check login of lighting come " repeating historical ".Batch mode and with the difference between the reserve that logs on as the basis be the former backing up be compatible and illustrated be the state of its last writing time, login and database then in order to show the compatibility on the time point, must further be handled in event of failure.
The people's such as Gawlick that on March 26th, 1989 issued United States Patent (USP) 4,507,751 " using the method and apparatus of the log data of leading write data collection " illustration a transaction management system, wherein all affairs are to be recorded in the daily record on the basis of leading write data collection.In this patent, a working cell is at first to be recorded in reserve medium (daily record) to go up and then write on its external storage address.
The appended document of carrying out in the maintenance of DASD medium generates
The people's such as Anglin that submit to May 16 nineteen ninety common pending application USSN07/524,206(IBM list of references SA9-90-012) " carrying out the critical method and apparatus of taking inventory command fetch " instructed with to the active access of the other parts of DASD medium concurrent on the selection part of the magnetic medium of the operation capable of circulation in a shunting, carry out the medium maintenance.The method of Anglin requires customer data in a target track to for moving with the segmentation between the road, and the access request that all are concurrent is distributed to one or more for on the road, and from safeguarding and return and duplicate for finishing to target track with the road.
The request and the interruption that occur before the customer data of carrying out the road moves cause process to be reset.In another case, the request that occurs in the execution data move is then seen a device busy (DEVICE BUSY) state with interruption.What this caused asking etc. requeues.
An object of the present invention is to design a kind of method and apparatus, be used on external storage, backing up record consistently, and this duplicating is and duplicates concurrent the carrying out of pause that caused very short CPU application program is carried out by described.
An associated purpose is that design a kind ofly can be supported completely, reserve scheduling strategy increment or that mix back up method and apparatus.
Above-mentioned purpose is to rely on the data map that will duplicate on the atom level to backing up on the medium and using appended document to come buffer-stored to be reached by a kind of method and apparatus of concurrent any data subset that renewal influenced.To upgrade concomitantly under this condition that allows to keep compatibility at the same time and duplicate order and write on the external storage fully.
Method of the present invention is to realize by the data set of backing up appointment in the dialogue that identifies uniquely.Each dialogue comprises that dialogue login and initialization also copy to a predetermined instant (t simultaneously
0) till the state of data designated collection, and with t
0All later renewals are all write on the external storage.This method comprises the steps: (1) with the influenced of data set but the part of not duplicating is as yet write as appended document, and (2) upgrade raw data on the spot on described external storage, and (3) copy to appended document on the medium to back up order.
The integrality of the data set that advantageously, is replicated is kept and the dead time of process exists hardly.And different with the documents of above-mentioned Anglin, method and apparatus of the present invention is to be target with a kind of new purposes that appended document generates.That is, the difference between them is that wherein, the use of appended document had both made things convenient for reserve data set in the normal replication order, had made things convenient for the stack of reserve and renewal again in the use of generation of appended document of the part of not duplicating of a data set.
Fig. 1 has showed a typical multiple processing multiprogramming environment according to prior art, wherein executory process with use randomly or sequentially from the external storage access data.
Fig. 2 shows the time line chart criticizing or flow to the reserve window in the journey according to prior art.
The reserve window that Fig. 3 has described as almost eliminating of the result of method and apparatus of the present invention.
Fig. 4 has shown t of the present invention
0Back up the notion flow process of method.
Fig. 5 and 6 is illustrated respectively in the control flow on external storage control and the CPU operating system grade.
Carry out the exemplary CPU environment of method of the present invention
The present invention can carry out in following configuration easily, and each CPU in one of them system can have a band IBM MVS(multiple virtual storage as an example) IBM/360 of operating system or the cpu type of 370 architectures.The people's such as Amdahl that issue in September 3 nineteen sixty-eight USP3 has fully described the CPU of IBM/360 architecture in 400,371 " data handling systems ".The people's such as Luiz that issue on June 10th, 1980 USP4 has described in 207,609 " are independent of in CPU more than and shared equipment access system the equipment in path keep reconnect " and has comprised a kind of configuration that CPU gets the coexistence of external storage.
At IBM communique GC28-1150, in " MVS/ extended architecture (XA) system library: the grand and instrument of system " volume MVS operating system has been described also.Call and task charges to and wait for that the MVS of these standards or the details of other operating system service omitted here such as the subsystem of partial blockade management, interruption or monitor.These OS service believes it is the fully understanding of those skilled in the art institute.
Arrive data the path, batch and interactive mode and backing up
Referring now to Fig. 1,, a multiple processing, multiprogrammed system have been described among Fig. 1 according to prior art.This system comprises channel request by redundancy/reply a plurality of processors (1,3) of interface (5,7,9) access external storage (21,23,25,27,29).As described in people's such as Luiz the patent, cpu process starts I/O to call a START I/O() in an IBM system 370 etc., set up the path that an arrival is stored in outside data by a MVS or other operating system, control is transferred to a channel subsystem, and this subsystem keeps the path of these data of arrival that transmit thereon.Usually, application program has data dependency and the operation that may pause is at short notice once fetched data or upgraded up to finishing.In transmission, this path is finishing up to transmission of being blocked.
Referring now to Fig. 2,, wherein shows the time line chart of criticizing or flow to the reserve window in the journey according to prior art.That is, in a lucky moment before reserve, application program is suspended or shuts down.Hang-up is till the reserve process is finished.Reserve finishes to represent to finish and submit to.What is called is finished and is meant that data that all requirements are duplicated are actually and reads from data source.The so-called submission to is meant that all data that will duplicate write on the output medium in fact.
Logic finished from physics finish separately
Referring now to Fig. 3,, wherein described according to basic elimination as the reserve window of a kind of structure of method and apparatus of the present invention.In case back of the present invention Preparation Method (with duplicating) process brings into operation, data (with regard to duplicating) " are freezed " on this point in the time.On in the time this point, duplicate that to be said to be " logically to have finished ".Submit state promptly " is physically finished " state and then will just be occurred after a while.On the point of " logically finishing " in the time, these data can be employed program fully and use once again.From sending t
oReserve the time be carved into the time of data between can being utilized again in low submicrosecond (sub-second) scope.In other words, total application data pauses (reserve window) time with millisecond meter.
Abnormal ending
If t
0The reserve process logic finish a little and physics finish a little between abnormal ending, then back up useless and process must restart, this on the one hand method and apparatus of the present invention have identical shortcoming with prior art.That is, all reserve all must be carried out again.A kind of restriction is the time criticality that has lost snapshot.
Concept characteristic
Referring now to Fig. 4 and 5,, show a conceptional flowchart of method of the present invention there.It is pointed out that each reserve dialogue is to be endowed a unique dialogue identifier (ID), and comprise an initialization and a reserve processing section.Because a plurality of reserve dialogue might concurrent running, thereby each is talked with ID its " snapshot " and is and inequality.
Each CPU comprises the operating system with memory manager program part.Usually, the 370 type CPU of IBM system that move under MVS operating system will comprise the memory manager program of data device data collection service (DFDSS) type, United States Patent (USP) 4 as the people such as Ferro that issued on August 8th, 1989,855, described in 909 " keeping the method to the mobile VSAM basic group of the alternate index of trooping ", DFDSS is to describe in the moon/day/year on IBM communique GC26-4388 " data equipment data set service: the users' guidebook " date also.
Data are logically to be organized into record and data set.The real address of data in external storage DASD(direct access storage device) volume, road, cylinder are represented.The virtual address of data is then represented with base address+side-play amount and scope.
A record can have counting-key word-data layout.It can occupy one or more unit of real storage.Data set can be stored on the linkage unit of real storage or can scatter as a logical collection of a plurality of records.The thing followed is, if reserve is carried out on one group of data set, that just is necessary to carry out multiple ordering to produce the inverted index to real storage.
For target of the present invention, reserve is handled and is managed on two-stage, that is, and and in CPU OS resource manager level (Fig. 1-1,3) and in memory control unit level (Fig. 1-2 1,23).
Referring to Fig. 4 and 5, initialize process comprises that one of response reception is duplicated or three big steps of the requested resource supervisory routine (for example DFDSS) of reserve particular data again.These steps comprise the sorting data collection, set up one or more bitmaps and a calling process on CPU sends logic and finishes signal.The path elements of data set data access that list or sign is ordered into the quantization degree in DASD road.Along with, set up the bitmap that data set is associated with access path, any in them comprises into given program of duplicating in the dialogue or being left out.At last, resource manager sends the signal that logic is finished, as long as just expression upgrade and can on this data set, handle after having passed through the delay of a short time.
More specifically, memory resource supervisory routine (DFDSS) receives one and duplicates or the request of reserve data.Usually, the form of this request is the table of a data set or the table through filtering of a data set.DFDSS is mapped to this request in the table of a physical extent with DASD storage volume and storage control unit (SCU).Then, DFDSS writes down this request with the SCU of each reference.At this moment, determined the ID of this dialogue and set up this dialogue.
Should be appreciated that DFDSS is that all scopes by each volume that will duplicate are sent to each SCU and come with each this dialogue of SCU initialization.Then, each SCU will set up a bitmap for each volume of respectively participating in this dialogue.Which road this bitmap will indicate is this t
0Duplicate the part of dialogue.Control turns back to DFDSS.This is that data can be used the point of " logically finishing " again.DFDSS correspondingly notifies such as the such operating system part of a scheduler program in the equipment of system management.
Reserve is handled
After the initialization, DFDSS begins to read the road of being asked.At t
0Duplicate and talk with when enlivening, each SCU monitors all renewals.If receive a renewal, SCU carries out one giving of taking into account of this renewal is decided algorithm.
If this renewal is at not at t
0A volume of dialogue, then this renewal is normally finished, otherwise, if this renewal is at a volume of a part of talking with, then check the bitmap to judge whether this road is shielded.If this position is that off(supposes that this is indicated as Binary Zero), this represents that current renewal the in this road normally finish not in duplicating dialogue.Meaningfully, if this road is shielded (this position is on), this represents that this is to duplicate the part of dialogue and it is not read by DFDSS as yet together.In this case, SCU
(1) keeps this renewal.
(2) independently (this comprises t together in the caches part this one slave unit to be separated to one
0The reserve process existing data zero hour).
(3) allow renewal to proceed.
(4) if independently comprise any road in the caches part at one, DFDSS promptly reads these roads to minimize the influence of normal cache operation.
Referring to Fig. 4, wherein described the step of this method again.In this figure, the renewal in road 4 and road 7 is caused unaltered road is split into before upgrading independently in the caches part finishing, subsequently, DFDSS reads these roads from caches independently partly.Still unripe those roads that merge on the output medium that DFDSS reads are temporarily to be stored in the main appended document.
Caution signal is handled and is used to guarantee the independently capacity of the unlikely excessive consumption cache memory of caches part.When a caution signal was appeared in one's mind on the main frame, operating system was by a DFDSS, and the latter empties independently caches part immediately.
In Fig. 4, locate to occur to t at " A "
0The application that the data that duplicating process duplicated are carried out is at random upgraded.The original map in these roads is copied to independently in the caches part.DFDSS locates to read unaltered road from DASD equipment at " B ".If any road is at t
0Process initiation is changed later on.Then they do not return to DFDSS.When blocking the way was located to be moved to as the consequence of upgrading independently in the caches part at " C ", a threshold value was noted interrupting appearing in one's mind on the main frame.These interruptions are served by operating system.Operating system is sent appropriate command to obtain the reason of interruption to SCU.If this interruption is to a specific t
0Process, then this indication is passed on the DFDSS.
In case DFDSS receives this interruption, it just begins to empty the road that is accumulated in the part of caches independently.Any having read and the still unripe road that is placed on the output medium are considered to " breaking away from sequence " and temporarily are stored in the primary memory appended document.
As last measure, direct reading of data and will be stored in the data that main appended document accounts for and finally merge on the output medium from the DASD equipment with suitable sequence in step " D ".
Referring to Fig. 4 and 5, suppose that is called this t once again
0The process of process requires to back up to be stored in 100 data sets that give on the fixed DASD road.If neither one is changed in these roads in duplicating process, DFDSS can be read-only by way of 1-100 and place them on the output medium.In order to allow concurrent renewal external storage when backing up, must suppose simultaneously that being stored in one or more data that these give on the fixed road has the rational possibility of being changed.
Suppose that this process has begun and DFDSS has duplicated road 1-20.This means that it also will duplicate 21-100.If one is used or process attempts to change 7, this allows that " as usual " finish, because road 7 was replicated.Yet, if attempt to change 44, just this change can not " as usual " finish because road 44 is not replicated as yet.Be necessary to guarantee that saying 44 to duplicating keeps its original state.So upgrading before the road that does not duplicate, allow to finish change before in appended document of an interim duplicate existence with road 44.This together interim duplicate of 44 be arranged in one independently the caches part retrieve later on for DFDSS.DFDSS retrieves this one and carves in due course road 44 is placed on the output medium.
The reserve process makes DFDSS obtain being stored in the data of giving on deciding from two sources:
(1) directly from road that DASD reads.These are at t
0(by an application) road of changing not after duplicating process begins.
(2) from road that caches partly reads.These are at t
0The original map in the road that changed after process begins.
Because a purpose is the influence that minimizes normal cache operation, the road is once reading in independently caches part, and they just can read for DFDSS.
The detailed logic flow process that reserve is handled
Referring now to Fig. 5 and 6,, show some process flow diagrams there.Fig. 5 has comprised initialization and SCU reserve is handled, and the CPU OS that Fig. 6 has then described appended document handles (asynchronous process) and the CPU OS management (synchronous processing) of duplicating the dialogue data transmission from SCU to the output medium.For integrality, in this section, replenish these and describe with the more detailed flow process of control inventory.These inventories are a kind of many-one mappings to the process flow diagram of being described in Fig. 5 and 6.
Initialization flow process inventory
Initialize process starts when CPU operating system (OS) is received a reserve or duplicate the request of data of some.Following logical process is pressed in this request:
1. set up the table of the data set of wanting reserve.
2. roll up with the DASD at data set place data set table is sorted.
3. find out which volume and belong to which SCU.
4. each SCU in the notification dialog and set up a dialogue identifier unique on all SCU.
5. for each volume on each SCU, notifying which road is the part that this To duplicates dialogue.
A.SCU sets up a bitmap with each volume that is in this dialogue
B. in this bitmap, one " 0 " represents that this road is not the part that this To duplicates dialogue.The corresponding road of one " 1 " expression is the part that this To duplicates dialogue.
A sign indication arrives " logic is finished " point and these data can be freely used in application again 6.CPU OS returns to calling process.
SCU flow process inventory
This stage comprises two processes of carrying out simultaneously, is carried out by SCU for one, is carried out by CPU operating system for one.SCU
1. for the renewal of each appearance, once check to judge that whether this renewal has a volume that a TO duplicates dialogue at current.
2., upgrade and normally finish if the answer to #1 is not.
If to the answer of #1 for being that the bitmap of correspondence is tested to judge whether a road that upgrades is the part that TO duplicates dialogue.
4., then upgrade and normally finish if the answer to #3 is not.
If to the answer of #3 for being then to carry out following step:
A. temporarily keep this renewal.
B. the road that will will upgrade at once copies in the appended document district in this SCU cache memory.
C. allow and finish this renewal.
D. this road of bitmap item shutoff indication with this road no longer is the part that TO duplicates dialogue.Therefore, later renewal is unrestricted.
E. check the current road number that is included in the appended document whether to surpass one and give fixed threshold value.
(1) if it is no more than this threshold value, then proceeds.
(2) if having surpassed this threshold value, it proposes to notice that indication must be read appended document (emptying) immediately to CPU OS.
6. any (from the DASD) that only occurs in the TO duplicating process in CPU OS reads and causes taking following step:
A. the data track of being asked is sent to CPU OS TO duplicating process.
B. indication this road with regard to this SCU, corresponding position that turn-offs in the bitmap no longer is the part that this TO duplicates dialogue.
7. when all positions in all bitmaps among (belonging to an a single dialogue) SCU all were turned off, for this SCU, this dialogue had been finished in fact.
CPU OS flow process is made of an asynchronous process and a synchronized process.
Asynchronous process
1. wait and listen an attention (delivering to any " signal " that gives the generation of fixed incident of an indication of CPU OS from a SCU).
2. when on a SCU, occurring noting, begin to be read sky up to this appended document from this SCU appended document reading of data.
3. all be the road of " disengaging sequence " from each road that this appended document reads, and be stored in the main working document and be ready to be placed on the output medium up to it.
4. to #1.
Synchronized process
Remember this point, t
0Duplicating process is that the order with appointment begins the reading of data road.
1.TO duplicating process determines will read which road in a single I/O request.
2. inquire about main working document to judge whether arbitrary road that will read is in this working document.
If A. the answer to #2 is not, suppose still that then this road is present on the DASD equipment with unaltered state.
If B. to the answer of #2 for being then to change this reading order to avoid reading again a road of having read.That is, this road had been updated and original road has been split in the appended document and has finally moved on to main working document in front.
3. send the dialogue in some roads is read
If A. SCU indication is attempted a dialogue to a current not road in this dialogue and read, CPU OS supposes that this road is arranged in the SCU appended document or main working document also recovers this road therefrom.
4. the data that obtain from #3 are write on the output medium after merging with any data track that obtains from step #2B.
5. if the road that will read in addition is to #1.
Otherwise, when all roads have all read and have write output medium:
A. finish this dialogue and SCU that all participate in.
B. return " physics an is finished " signal to calling process.This indication wants the data of reserve in fact to write on the output medium.
Extend
Though described the present invention in the context of an IBM MVS operating system, it can be implemented in any general-purpose operating system of buying too, such as the VM(virtual machine), OS2(system 2) etc.Equally, though the DFDSS that named as an exemplary external storage resource manager, the present invention can operate and undue experimentation work that need not those of ordinary skill with any suitable supervisory routine.
These and other extension of the present invention can be made not breaking away under spirit that appending claims states and the scope.
Claims (8)
1, the method that realizes of a kind of CPU is used for backing up the data designated collection that embodies the time point data compatibility on a memory sub-system that is connected to described CPU, and described backing up is and the application execution concurrence of CPU to it is characterized in that following step:
(a) hang up to use carry out and the logic that constitutes a data set to physical storage subsystem allocation index, and recover to use after this and carry out;
(b) by with the data designated collection from memory sub-system copy to CPU and by the memory sub-system position that CPU writes other with them come a kind of give on the basis fixed or at random with the data designated collection physically reserve on this memory sub-system; And
(c) by following step on this memory sub-system, handle to the specific data collection that do not duplicate by the described renewal of any renewal buffer-stored of using starting, write by these and upgrade the data set influenced or the appended document of its part, this memory sub-system is directly write in these renewals, and appended document copied to CPU and with the defined order of backing up of index the specific data collection is write on the storer in the mode with step (b) on the CPU.
2, the method that realizes of a kind of CPU is used for being backlogged the pause that minimizes when duplicating the operation that is positioned at a described data set on the external storage that is connected to a CPU when data set, it is characterized in that following step:
(a) thus giving fixed data set scope with one on atom level copies to a second place scope from a primary importance scope of described external storage and sets up one in described second scope and back up product; And
(b) by following step process to being positioned at any concurrent renewal of any part of not duplicating of giving given data collection scope on first scope:
(1) will be upgraded the part of not duplicating that influences and be write on the appended document,
(2) on the position of first scope, handle these renewals, and
(3) with the common order of backing up appended document is write on second scope.
3, according to the method for claim 2, it is characterized in that described method also comprises step (c) if upgrading that part of of data set scope that is influenced has been replicated, then handle renewal on the data set scope in first scope unlimitedly.
4, a kind of method, be used to manage with the data set of application execution concurrence on a CPU and back up, this CPU is the external subsystems of the circulation memory device in a shunting of connection, liaison ground on the access path of a definition at least, it is characterized in that following step:
(a) hang up the data set that the application on described CPU is carried out and specified at least one to be used to back up, and request is from a duplicate of the described specific data collection of storage subsystem;
(b) logic backed up of composition data collection on storage subsystem and equipment road index subsequent signal notice CPU is finished;
(c) the described logic of response is finished signal recovers to use execution on described CPU;
(d) the specific data collection is copied to CPU from storage subsystem giving on the basis fixed or at random;
(e) directly write on the storage subsystem (unless they are at a part of a specific data collection that does not copy to CPU) handles any application of the data set in the storage subsystem is upgraded by upgrading, otherwise buffer-stored should be upgraded, write the appended document that does not duplicate part of a specific data collection, to upgrade directly then and write on the storage subsystem, and appended document will be copied to CPU; And
(f) with step (d) and (e) asynchronous duplicating on the order, CPU writes other storage subsystem position with the specific data collection and the appended document of accumulation.
5, according to the method for claim 4, the index that it is characterized in that data set and position, road represents with bitmap, the specific data collection that does not wherein duplicate be with one first Boolean sign duplicate with unspecified data set then with one second Boolean sign; And wherein when a specific data collection with its equipment road content representation is copied to CPU, the bitmap sign that it is corresponding changes to the step of second Boolean from first Boolean.
6,, it is characterized in that this method also comprises the steps: according to the method for claim 4
(g) only when all specific data collection and appended document have been write described other storage subsystem position, the physics that the ability signaling is backed up is finished.
7, according to the method for claim 4, it is characterized in that step (e) can revise make appended document be accumulate on the storage subsystem and only when a number of thresholds occurring, just copy to CPU.
8, a system of a CPU formation of the position data set from the external subsystems of quoting the circulation memory device that is connected a shunting on the CPU or its part is characterized in that:
(a) be used for giving the device that the given data collection copies to a second place scope (backing up) atomically with one from a described outside primary importance scope of storing up on the subsystem; And
(b) be used for the device of following step process any concurrent renewal of being positioned at any part of not duplicating of giving given data collection scope on first scope:
(1) will be upgraded the part of not duplicating that influences and be write on the appended document,
(2) on first range position, handle renewal, and
(3) to back up order normally appended document is copied on second scope.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US78104491A | 1991-10-18 | 1991-10-18 | |
US07/781,044 | 1991-10-18 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1071770A true CN1071770A (en) | 1993-05-05 |
CN1025381C CN1025381C (en) | 1994-07-06 |
Family
ID=25121497
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN92110772A Expired - Fee Related CN1025381C (en) | 1991-10-18 | 1992-09-18 | Method and means for time zero backup copying of data |
Country Status (6)
Country | Link |
---|---|
EP (1) | EP0608255A1 (en) |
JP (1) | JPH05210555A (en) |
KR (1) | KR950014175B1 (en) |
CN (1) | CN1025381C (en) |
CA (1) | CA2071346A1 (en) |
WO (1) | WO1993008529A1 (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1293461C (en) * | 1999-07-30 | 2007-01-03 | 神基科技股份有限公司 | Method for suspending state of computer system |
CN105960635A (en) * | 2014-02-07 | 2016-09-21 | 国际商业机器公司 | Creating restore copy from copy of source data in repository having source data at different point-in-times |
CN108228647A (en) * | 2016-12-21 | 2018-06-29 | 伊姆西Ip控股有限责任公司 | For the method and apparatus of data copy |
US10176048B2 (en) | 2014-02-07 | 2019-01-08 | International Business Machines Corporation | Creating a restore copy from a copy of source data in a repository having source data at different point-in-times and reading data from the repository for the restore copy |
US10387446B2 (en) | 2014-04-28 | 2019-08-20 | International Business Machines Corporation | Merging multiple point-in-time copies into a merged point-in-time copy |
US11169958B2 (en) | 2014-02-07 | 2021-11-09 | International Business Machines Corporation | Using a repository having a full copy of source data and point-in-time information from point-in-time copies of the source data to restore the source data at different points-in-time |
US11194667B2 (en) | 2014-02-07 | 2021-12-07 | International Business Machines Corporation | Creating a restore copy from a copy of a full copy of source data in a repository that is at a different point-in-time than a restore point-in-time of a restore request |
Families Citing this family (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5675725A (en) * | 1993-07-19 | 1997-10-07 | Cheyenne Advanced Technology Limited | Computer backup system operable with open files |
ATE176824T1 (en) * | 1993-07-19 | 1999-03-15 | Cheyenne Advanced Tech Ltd | FILE BACKUP SYSTEM |
JPH07210429A (en) * | 1994-01-11 | 1995-08-11 | Hitachi Ltd | Dump acquiring method, controller and information processing system |
GB2290396A (en) * | 1994-07-20 | 1995-12-20 | Intelligence Quotient Int | Backing-up shared data |
WO1997024667A1 (en) * | 1995-12-28 | 1997-07-10 | Ipl Systems, Inc. | Dasd storage back-up using out-of-sequence writes |
US6081875A (en) * | 1997-05-19 | 2000-06-27 | Emc Corporation | Apparatus and method for backup of a disk storage system |
JPH10333961A (en) * | 1997-05-30 | 1998-12-18 | Kokusai Zunou Sangyo Kk | Data base recovery system and computer-readable recording medium recording recovery program |
US7039657B1 (en) * | 1999-11-09 | 2006-05-02 | International Business Machines Corporation | Method, system, and program for accessing data from storage systems |
JP3714184B2 (en) | 2001-03-29 | 2005-11-09 | 富士通株式会社 | Copying method between data areas of storage device and storage system |
US7213103B2 (en) | 2004-04-22 | 2007-05-01 | Apple Inc. | Accessing data storage systems without waiting for read errors |
US8495015B2 (en) | 2005-06-21 | 2013-07-23 | Apple Inc. | Peer-to-peer syncing in a decentralized environment |
US7523146B2 (en) | 2005-06-21 | 2009-04-21 | Apple Inc. | Apparatus and method for peer-to-peer N-way synchronization in a decentralized environment |
JP4414381B2 (en) | 2005-08-03 | 2010-02-10 | 富士通株式会社 | File management program, file management apparatus, and file management method |
US7797670B2 (en) | 2006-04-14 | 2010-09-14 | Apple Inc. | Mirrored file system |
JP2008027163A (en) * | 2006-07-20 | 2008-02-07 | Fujitsu Ltd | Data recorder, data recording program, and data recording method |
US7860826B2 (en) | 2006-08-04 | 2010-12-28 | Apple Inc. | Method and system for using global equivalency sets to identify data during peer-to-peer synchronization |
US7657769B2 (en) | 2007-01-08 | 2010-02-02 | Marcy M Scott | N-way synchronization of data |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4686620A (en) * | 1984-07-26 | 1987-08-11 | American Telephone And Telegraph Company, At&T Bell Laboratories | Database backup method |
US4752910A (en) * | 1984-10-30 | 1988-06-21 | Prime Computer, Inc. | Method and apparatus for continuous after-imaging |
JPS62203248A (en) * | 1986-03-03 | 1987-09-07 | Nec Corp | Dynamic saving and restoration system for data base |
JPH0290341A (en) * | 1988-09-28 | 1990-03-29 | Hitachi Ltd | Saving system for data base file |
US5027269A (en) * | 1989-04-27 | 1991-06-25 | International Business Machines Corporation | Method and apparatus for providing continuous availability of applications in a computer network |
US5454099A (en) * | 1989-07-25 | 1995-09-26 | International Business Machines Corporation | CPU implemented method for backing up modified data sets in non-volatile store for recovery in the event of CPU failure |
-
1992
- 1992-06-16 CA CA002071346A patent/CA2071346A1/en not_active Abandoned
- 1992-08-10 JP JP4211913A patent/JPH05210555A/en active Pending
- 1992-09-16 EP EP92919444A patent/EP0608255A1/en not_active Ceased
- 1992-09-16 WO PCT/EP1992/002127 patent/WO1993008529A1/en not_active Application Discontinuation
- 1992-09-18 KR KR1019920016989A patent/KR950014175B1/en not_active IP Right Cessation
- 1992-09-18 CN CN92110772A patent/CN1025381C/en not_active Expired - Fee Related
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1293461C (en) * | 1999-07-30 | 2007-01-03 | 神基科技股份有限公司 | Method for suspending state of computer system |
CN105960635A (en) * | 2014-02-07 | 2016-09-21 | 国际商业机器公司 | Creating restore copy from copy of source data in repository having source data at different point-in-times |
US10176048B2 (en) | 2014-02-07 | 2019-01-08 | International Business Machines Corporation | Creating a restore copy from a copy of source data in a repository having source data at different point-in-times and reading data from the repository for the restore copy |
US10372546B2 (en) | 2014-02-07 | 2019-08-06 | International Business Machines Corporation | Creating a restore copy from a copy of source data in a repository having source data at different point-in-times |
CN105960635B (en) * | 2014-02-07 | 2021-02-09 | 国际商业机器公司 | System and method for maintaining source data in a repository |
US11150994B2 (en) | 2014-02-07 | 2021-10-19 | International Business Machines Corporation | Creating a restore copy from a copy of source data in a repository having source data at different point-in-times |
US11169958B2 (en) | 2014-02-07 | 2021-11-09 | International Business Machines Corporation | Using a repository having a full copy of source data and point-in-time information from point-in-time copies of the source data to restore the source data at different points-in-time |
US11194667B2 (en) | 2014-02-07 | 2021-12-07 | International Business Machines Corporation | Creating a restore copy from a copy of a full copy of source data in a repository that is at a different point-in-time than a restore point-in-time of a restore request |
US10387446B2 (en) | 2014-04-28 | 2019-08-20 | International Business Machines Corporation | Merging multiple point-in-time copies into a merged point-in-time copy |
US11630839B2 (en) | 2014-04-28 | 2023-04-18 | International Business Machines Corporation | Merging multiple point-in-time copies into a merged point-in-time copy |
CN108228647A (en) * | 2016-12-21 | 2018-06-29 | 伊姆西Ip控股有限责任公司 | For the method and apparatus of data copy |
CN108228647B (en) * | 2016-12-21 | 2022-05-24 | 伊姆西Ip控股有限责任公司 | Method and apparatus for data copying |
Also Published As
Publication number | Publication date |
---|---|
CA2071346A1 (en) | 1993-04-19 |
KR930008636A (en) | 1993-05-21 |
KR950014175B1 (en) | 1995-11-22 |
EP0608255A1 (en) | 1994-08-03 |
WO1993008529A1 (en) | 1993-04-29 |
JPH05210555A (en) | 1993-08-20 |
CN1025381C (en) | 1994-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1025381C (en) | Method and means for time zero backup copying of data | |
CN1149480C (en) | Highly available file servers | |
US5555371A (en) | Data backup copying with delayed directory updating and reduced numbers of DASD accesses at a back up site using a log structured array data storage | |
CN1147788C (en) | Computer system and file management method | |
JP6437656B2 (en) | Storage device, storage system, and storage system control method | |
US7047380B2 (en) | System and method for using file system snapshots for online data backup | |
CN1143209C (en) | Operating system rebooting method | |
US10248322B2 (en) | Memory system | |
CN100339834C (en) | Recovery from failures within data processing systems | |
EP0983548B1 (en) | Apparatus and method for backup of a disk storage system | |
CN1408083A (en) | Scalable storage architecture | |
CN1804810A (en) | Method and system of redirection for storage access requests | |
US20090077140A1 (en) | Data Recovery in a Hierarchical Data Storage System | |
CN109284069A (en) | A kind of distributed memory system and method for storing Backup Data | |
CN1524221A (en) | Systems and methods of information backup | |
CN1342280A (en) | Protocol for replicated servers | |
CN1662903A (en) | Distributed network storage system with virtualization | |
CN1801107A (en) | Data recovery method | |
US8443371B2 (en) | Managing operation requests using different resources | |
US20040054849A1 (en) | System and method for detecting data integrity problems on a data storage device | |
CN1629817A (en) | Copying method, copying method and block identification data acquisition method | |
CN1568512A (en) | Data storage drives in an automated data storage library | |
CN1276349C (en) | Method for mirror backup of cluster platform cross parallel system | |
JP6860722B2 (en) | Memory system control method | |
US20090077004A1 (en) | Data Recovery in a Hierarchical Data Storage System |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C19 | Lapse of patent right due to non-payment of the annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |