CN108664496A - Data migration method and device - Google Patents
Data migration method and device Download PDFInfo
- Publication number
- CN108664496A CN108664496A CN201710197702.7A CN201710197702A CN108664496A CN 108664496 A CN108664496 A CN 108664496A CN 201710197702 A CN201710197702 A CN 201710197702A CN 108664496 A CN108664496 A CN 108664496A
- Authority
- CN
- China
- Prior art keywords
- relation
- relation chain
- node
- task
- chain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
Abstract
The invention discloses a kind of data migration method and devices, belong to network technique field.This method includes:According to the calculating task daily record of former service cluster, multiple relation chains are obtained, calculating task daily record is used to record the incidence relation of calculating task and business datum in former service cluster, and each relation chain is used to indicate one group of calculating task and business datum with incidence relation;As unit of relation chain, by indicated by multiple relation chains business datum and calculating task migrated successively to destination service cluster;When being migrated based on any one relation chain, the calculating task indicated by relation chain that is not migrated in the multiple relation chains of normal operation.By that will have the business datum of incidence relation and calculating task to be indicated using a relation chain, so that during carrying out Data Migration, it still can be with the calculating task indicated by relation chain that normal operation is not migrated, to not interfere with the normal use of business indicated by the relation chain not migrated.
Description
Technical field
The present invention relates to network technique field, more particularly to a kind of data migration method and device.
Background technology
With network technical development, the business datum amount of various businesses increases constantly swift and violent, can reach PB
(Petabyte, petabyte) grade even more than the order of magnitude so that internet and information industry enter the big data epoch.Big
Data age, generally use carry out business datum storage, business processing and business by the service cluster that a large amount of servers form
Management.In practical applications, service cluster would generally be deployed in the same IDC (Internet Data Center, in data
The heart) in computer room.However, with the continuous growth of business datum, the scale of service cluster is also constantly expanding, and IDC computer rooms
Scale is limited, wherein the Servers-all for failing to lay down the service cluster may be deposited, to limit the scale of service cluster,
At this point, in order to meet the needs of business datum growth, it can be by the Data Migration in service cluster to larger new demand servicing
In cluster.
In the prior art, service cluster can be the corresponding calculating task of service creation and be this when carrying out business processing
The corresponding computing resource of distribution of computation tasks executes the processing procedure of business datum by running the calculating task.Due to each
Be typically to be mutually related between kind of business, in order to avoid when migrating the business datum of a business to other associated industry
Business impacts, it will usually the data of service cluster be carried out bulk migration and need first to stop all meters in bulk migration
After calculation task (stopping providing service to all business), all business datums are moved into new demand servicing cluster, then, are taken newly
Business cluster reconfigure calculating task and distribute corresponding computing resource, start the calculating task reconfigured later, to for
All business provide service again, to complete Data Migration.
In the implementation of the present invention, the inventor finds that the existing technology has at least the following problems:
Since the data volume of business datum in service cluster is huge, transition process usually require to spend several days, some months or
Person's longer time can cause all business all cannot normally make if stopping providing service to all business in this time
With.
Invention content
In order to solve problems in the prior art, an embodiment of the present invention provides a kind of data migration method and devices.It is described
Technical solution is as follows:
On the one hand, a kind of data migration method is provided, the method includes:
According to the calculating task daily record of former service cluster, multiple relation chains are obtained, the calculating task daily record is for recording
The incidence relation of calculating task and business datum, each relation chain are used to indicate with incidence relation in the original service cluster
One group of calculating task and business datum;
As unit of relation chain, by indicated by the multiple relation chain business datum and calculating task migrated successively to mesh
Mark service cluster;
Wherein, it when being migrated based on any one relation chain, is not migrated in the multiple relation chain of normal operation
Relation chain indicated by calculating task.
On the other hand, a kind of data migration device is provided, described device includes:
First acquisition unit obtains multiple relation chains, the calculating for the calculating task daily record according to former service cluster
Task daily record is used to record the incidence relation of calculating task and business datum in the former service cluster, and each relation chain is for referring to
Show one group of calculating task and business datum with incidence relation;
Migration units, for as unit of relation chain, the business datum indicated by the multiple relation chain to be appointed with calculating
Business is migrated successively to destination service cluster;
Wherein, it when being migrated based on any one relation chain, is not migrated in the multiple relation chain of normal operation
Relation chain indicated by calculating task.
The advantageous effect that technical solution provided in an embodiment of the present invention is brought is:
By according to the calculating task daily record in former service cluster, by business datum and calculating task with incidence relation
It is indicated using relation chain so that during carrying out Data Migration as unit of relation chain, the relation chain that is migrating
Other relation chains will not be had an impact, still can with the calculating task indicated by relation chain that normal operation is not migrated, from
The normal use of business indicated by the relation chain not migrated without influence.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, other are can also be obtained according to these attached drawings
Attached drawing.
Figure 1A is a kind of implement scene schematic diagram provided in an embodiment of the present invention;
Figure 1B is a kind of Organization Chart of migration platform provided in an embodiment of the present invention;
Fig. 2A is a kind of flow chart of data migration method provided in an embodiment of the present invention;
Fig. 2 B are a kind of relation chain schematic diagrames provided in an embodiment of the present invention;
Fig. 2 C are that a kind of relation chain provided in an embodiment of the present invention splits schematic diagram;
Fig. 2 D are that a kind of relation chain provided in an embodiment of the present invention splits schematic diagram;
Fig. 2 E are that a kind of relation chain provided in an embodiment of the present invention splits schematic diagram;
Fig. 2 F are that a kind of relation chain obtained by fractionation provided in an embodiment of the present invention accesses key business data
Schematic diagram;
Fig. 2 G are a kind of pair provided in an embodiment of the present invention and write the schematic diagram that table mechanism is related to flow;
Fig. 2 H are the transition state schematic diagrames in a kind of relation chain transition process provided in an embodiment of the present invention;
Fig. 3 is a kind of block diagram of data migration device provided in an embodiment of the present invention;
Fig. 4 is a kind of block diagram of data migration device provided in an embodiment of the present invention.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to embodiment party of the present invention
Formula is described in further detail.
Figure 1A is the implement scene schematic diagram of Data Migration provided in an embodiment of the present invention, referring to Figure 1A, the implement scene
Include former service cluster, destination service cluster and migration platform.
Wherein, former service cluster may include multiple storage clusters and multiple computing clusters, and storage cluster is for storing industry
The calculating for data, related data of the computing cluster for running calculating task and storage calculating task, such as the calculating task of being engaged in
The position etc. of resource size and computing resource.Wherein, storage cluster and computing cluster can be deployed in different services respectively
On device, it can also be deployed on identical server, the present embodiment is not construed as limiting this.
It should be noted that service cluster when carrying out business processing, can be the corresponding calculating task of service creation and be
The corresponding computing resource of the distribution of computation tasks executes one or more business processions by running the calculating task,
For example, some business datum is read from service cluster, by another business number of output after handling the business datum
According to write service cluster etc..Wherein, it can be a few houres, several that calculating task, which has certain cycle of operation property, the cycle of operation,
It, a few weeks or some months etc., such as calculating task that the cycle of operation is 1 hour is primary per operation every other hour.Wherein, no
The cycle of operation with calculating task can be the same or different, and be had by the type of calculating task and the processing speed of business datum
It closes, the present embodiment is not construed as limiting this.
In addition, also safeguarding there is data path mapping table in service cluster, which is used for business datum mark
Know the correspondence between the store path of business datum.Calculating task can be mapped by the data path in service cluster
Table determines the store path for the business datum for reading or being written, to according to the reading of the store path finishing service data of acquisition
It takes or ablation process.Wherein, the business datum that a calculating task is read can be written by other calculating tasks, and one
The business datum of calculating task write-in can be read again by other calculating tasks, in this way, between calculating task and business datum just
It is provided with certain input/output relation.
Wherein, migration platform is for migrating the data of service cluster, and is managed to data migration process, should
Migration platform can be deployed in former service cluster, can also be deployed in destination service cluster, it is of course also possible to be deployed in original
On other servers that can be communicated with the two other than service cluster and destination service cluster.In the present embodiment, migration
Platform is needed the Data Migration in former service cluster to destination service cluster, and the data of migration are related to the industry in former service cluster
Data of being engaged in and calculating task.
Specifically, which may include multiple modules, and each module plays different in data migration process
Effect.It is the Organization Chart for migrating platform referring to Figure 1B, which includes multiple function modules, below to each function module
Effect be introduced:
Wherein, analysis module is multiple according to calculating task log acquisition indicated by following step 201 to 203 for executing
The process of relation chain;It splits module and is used to execute the process that the relation chain indicated by following step 204 is split;Correction verification module is used for
Execute the process of the consistency desired result in following step 206 to migration subtask and relation chain.
Wherein, transferring module is for executing business datum migration involved in following step 205 to 208 and calculating task migration
Process, wherein by indicated by relation chain business datum complete migration after, transferring module execution data path mapping table
Store path handoff procedure, the process correspond to step 207.Wherein, the transition process of calculating task refers to being configured to calculating task
The configuration information of the handoff procedure of information, calculating task can be obtained from repository, which corresponds to step 208.Wherein, such as
The relation chain of fruit migration is that the relation chain obtained by fractionation then needs to synchronize key business data, which corresponds to step
Step a under rapid 206.
Wherein, it refers to that will migrate to the target storage of the business datum in destination service cluster that data path mapping table, which synchronizes,
Path is added in map paths table.
Wherein, migration platform foreground can be used for being managed the transition process of relation chain, for example can show relationship
The various information of chain, for example connection relation, the relation chain transition state residing in transition process of each node in relation chain, close
The nodal information of the running state information etc. of nodal information and calculating task in tethers, wherein relation chain includes in relation chain
Calculating task mark indicated by store path and task node indicated by all back end, wherein back end and task
The relevant explanation of node is referring to content shown in step 203.User can be by the migration platform foreground initiation or pause to closing
The transition process of tethers.
Wherein, repository be used for store calculating task computing resource configuration information, such as the size of computing resource with
And location information, in the repository can with storage service data former service cluster former store path and destination service collection
The target store path of group.Task nexus chain is for storing the multiple relation chains generated by analysis module.Migration task library is used for
The information of storage migration subtask, such as migration the numbering of subtask, the former store path of indicated business datum, business datum
With target store path, the information such as data volume size of business datum.
In one embodiment, former service cluster can be both placed in IDC1 computer rooms, and destination service cluster can be equal
It is placed in IDC2 computer rooms, IDC1 computer rooms are different with the geographical location where IDC2 computer rooms.The wherein scale of destination service cluster
More than the scale of former service cluster, the number that corresponding IDC2 computer rooms can accommodate server can accommodate server more than IDC1 computer rooms
Number.Certainly, in another embodiment, former service cluster or destination service cluster can be prevented in different IDC machines
Fang Zhong, the present embodiment are not construed as limiting this.
Fig. 2A is a kind of flow chart of data migration method provided in an embodiment of the present invention, and referring to Fig. 2A, the present invention is implemented
Example provide method flow include:
201, the calculating task daily record of former service cluster is obtained.
The calculating task of former service cluster in the process of running, can generate calculating task daily record, the calculating task day
Will is used to record the incidence relation of calculating task and business datum in former service cluster.For example, the calculating task daily record includes meter
The task identification of calculation task, the store path of business datum, calculating task and business datum input/output relation and other
Information, the other information may include the business information belonging to calculating task, such as the user information belonging to service identification, business
Deng.Wherein, the store path of business datum can be used to refer to business datum, and identical store path is used to indicate identical industry
Business data, calculating task access the business datum by the store path of business datum.
Migration platform can obtain the calculating task daily record from former service cluster, and be extracted from the calculating task daily record
A plurality of input-output record.Wherein, input-output record be used to indicate the task identification of calculating task, business datum storage road
The input/output relation of diameter and calculating task and business datum is as shown in table 1 a kind of input-output record table.
Table 1
In the present embodiment, migration platform can carry out a plurality of input-output record extracted from calculating task daily record
Analysis is used to indicate calculating task and multiple relation chains of business datum incidence relation to obtain, the multiple relation chains of the acquisition
Process step 202 as described below is to 204.
202, a plurality of input-output record recorded according to calculating task daily record, for the input and output with incidence relation
Record adds identical relation chain mark, and different relation chain marks is added for the input-output record without incidence relation.
In the present embodiment, migration platform is that the input-output record with incidence relation adds identical relation chain mark
Process can be:Every input-output record in a plurality of input-output record is traversed, for currently traverse first
Input-output record, if the input-output record traversed includes having incidence relation between the first input-output record
The second input-output record, then be that the first input-output record adds identical with the second input-output record relation chain mark
Know;If not including second defeated with incidence relation between the first input-output record in the input-output record traversed
Enter output record, then adds the relation chain mark different from the input-output record traversed for the first input-output record.
Wherein, refer to the first input with incidence relation between the first input-output record and the second input-output record
The indicated calculating task of output record has input/output relation with the business datum indicated by the second input-output record, or
Person, the business datum indicated by the first input-output record have input with the calculating task indicated by the second input-output record
Output relation.
For example, by taking table 1 as an example, each input-output record of table 1 is traversed, when first input and output of traversal
When record, a relation chain mark 1001 is added for the input-output record, it is assumed that the first input-output record currently traversed
For Article 2 input-output record " task 2, IN, store path 1 ", then due to task 2 and first input and output traversed
" store path 1 " has input relationship in record, it is determined that the input-output record traversed include with currently traverse the
One input-output record has the second input-output record of incidence relation, then is first input-output record addition and first
The identical relation chain mark of input-output record 1001.Assuming that the first input-output record currently traversed is last in table 1
One input-output record " task 5, IN, store path 5 ", by " task 5 " and all input-output record institutes traversed
It indicates between business datum there is no input/output relation, and " store path 5 " and all input-output records for having traversed
There is no input/output relations between indicated calculating task, therefore, in the input-output record traversed do not include second defeated
Enter output record, therefore, is added for the last item input-output record currently traversed and traversed input-output record not
Same relation chain mark, such as different relation chain mark can be with 1002 etc..Relationship is being added for all input-output records
After chain mark, relationship chained list as shown in Table 2 can be obtained.
Table 2
203, between the calculating task and business datum indicated by the input-output record according to identical relation chain mark
Incidence relation generates multiple first relation chains.
In the present embodiment, the input-output record identified with identical relation chain can be abstracted into one by migration platform
A first relation chain, the first relation chain include the task node for being used to indicate calculating task, the number for being used to indicate business datum
According to the input/output relation between node and task node and back end.Wherein, the task node packet in the first relation chain
The task identification of calculating task is included, back end includes the store path of business datum.
By taking table 2 as an example, according to the input-output record with identical relation chain mark 1001, the first relation chain of generation is such as
Show that relation chain mark 1001 corresponds to indicated by input-output record between business datum and calculating task in Fig. 2 B, Fig. 2 B
Incidence relation, first relation chain include the task node 1-4 corresponding to task 1 to 4, the data corresponding to store path 1 to 4
Input/output relation between node 1-4 and service node and back end.Wherein, back end 1 is directed toward by task node 1
Line be used to indicate calculating task 1 to store path 1 be written business datum.The line of task node 2 is directed toward by back end 1
It is used to indicate calculating task 2 and reads business datum from store path 2.
In the present embodiment, business datum or calculating task indicated by the first different relation chains, with other the first relation chains
Do not have incidence relation between indicated calculating task or business datum.It therefore, can be as unit of the first relation chain, to original
Business datum and calculating task in service cluster are migrated, and are appointed with calculating to the business datum indicated by a relation chain
When business is migrated, the normal operation of calculating task indicated by other first relation chains is not interfered with.
In view of the time of Data Migration can be by the double constraints of the data volume and network bandwidth that migrate, and usual network
Bandwidth is limited, in order to guarantee to complete the migration of data indicated by a relation chain within a short period of time, thus into one
Step reduces influence of the transition process to business normal use, and the present embodiment can further close corresponding data amount larger first
Tethers is split, and detailed process is referring to step 204.
If 204, multiple first relation chains include the second relation chain, the second relation chain is split as multiple thirds and is closed
Tethers, the second relation chain are that the data volume of indicated business datum is more than the first relation chain of first threshold.
Wherein, first threshold can be set by migration platform according to the default transit time and network bandwidth of relation chain, example
Such as, it is assumed that network bandwidth is 2GB/s (giga bits per second), and it is 2 minutes to preset transit time, then first threshold is up to 120GB,
Certain first threshold might be less that the 120GB, be impacted to network bandwidth to avoid since network environment is unstable.Its
In, default transit time can be preset by migration platform, or be set etc. according to the business demand of user, this
Embodiment is not construed as limiting this.
For the first relation chain of each of multiple first relation chain, migration platform can be according in first relation chain
Store path indicated by back end obtains the data volume of business datum indicated by first relation chain.If first pass
Business datum indicated by tethers is more than the first threshold, it is determined that first relation chain is the second relation chain, and determines needs pair
Second relation chain is split, and the process of the fractionation may comprise steps of 204a to 204c:
Step 204a, the weights of multiple back end in the second relation chain are obtained.
Wherein, the weights of each back end are used to indicate correlation degree of the back end in the second relation chain, weights
More high associated degree is higher.
The acquisition process of the weights of multiple back end can be:For each data section in multiple back end
Point, by the product of the data volume of business datum indicated by the number of task node associated with back end and back end,
It is determined as the weights of back end.
By taking the first relation chain shown in Fig. 2 B as an example, wherein task node associated with back end 1 includes task section
The number of point 1 to 4, task node is 4, it is assumed that the data volume of business datum indicated by the back end is 100GB, then the data
The weights of node are that 4*100 is equal to 400.
It should be noted that since the purpose that the first relation chain is split is that the larger relation chain of data volume is split as data
The smaller relation chain of amount, and for any data node in relation chain, if task node associated with the back end
It is more, then show that the number for the relation chain that can be split based on the back end is more, so just so that fractionation obtained
The data volume of business datum indicated by each relation chain is more balanced, and the data volume of some relation chain will not be caused excessive, therefore,
It needs to consider indicated by number and the back end of task node associated with back end in the weights for determining back end
The two factors of the data volume of business datum.
Step 204b, the position of sequence and multiple back end in the second relation chain according to weights from high to low, from
Critical data node is obtained in multiple back end, which can tear the second relation chain open for first in sequence
It is divided into the back end of at least two third relation chains.
In the present embodiment, in order to improve relation chain fractionation efficiency and success rate, migration platform according to weights from height to
Low sequence analyzes each back end, is carried out in advance for example, migration platform is based on second relation chain of back end pair
It splits, is determined to splitting into the second relation chain into the number of third relation chain, if splitting the number of obtained third relation chain
Mesh is less than 2, then the putting in order from high to low according to weights, and the process split in advance is executed to next back end;If torn open
The number for the third relation chain got is not less than 2, then the back end is determined as critical data node, is based on the key number
It is split according to the second relation chain of node pair.After getting critical data node from multiple back end, migration platform is not
The process split in advance is executed to the back end after the critical data node in above-mentioned put in order again.
Wherein, it when being split in advance according to the second relation chain of back end pair, is determined to split obtained third pass
The method of the number of tethers can be:Disconnect the association between the back end and task node associated with the back end
Relationship determines between the node (including task node and back end) in the second relation chain in addition to the back end later
Connectivity, if between node in addition to the back end being connection, it is determined that the third relationship that can be split
The number of chain was 1 (being less than 2), otherwise, it determines the number for the third relation chain that can be split is not less than 2.
Wherein it is determined that the process of the connectivity between node in the second relation chain in addition to the back end can be:
Node in addition to the back end is traversed, for example, can an optional node be starting point traversed, if each
Node can traverse, it is determined that the node in addition to the back end is connection, otherwise, it determines in addition to the back end
Node be disconnected.
It should be noted that the above-mentioned process split in advance to the second relation chain is not to carry out reality to the second relation chain
The process of fractionation, but migrate platform and assume that the second relation chain can be split into how many a thirds based on corresponding data node closes
The analytic process of tethers.
Step 204c, it is based on the critical data node, it will be associated more with the critical data node in the second relation chain
A task node is split into multiple third relation chains.
In the present embodiment, migration platform is based on the critical data node, and the second relation chain is split as multiple third relationships
The process of chain can be divided into following three kinds of situations:
The first situation, for each task section in the multiple tasks node that is directly linked with the critical data node
Point, when the critical data node, the task node and the critical data node and the task node are disconnected relationship with
There is the task node node of incidence relation to be determined as third relation chain.
In that case, include the critical data node in each third relation chain.It will again be assumed that shown in Fig. 2 B
Critical data node is back end 1 in relation chain, as shown in Figure 2 C to be based on back end 1 in the case of this kind, to Fig. 2 B institutes
The schematic diagram for multiple third relation chains that the relation chain shown is split.
The critical data node is determined as a third relation chain by the second situation, for the critical data node
Each task node in associated multiple tasks node, by it is in addition to the critical data node and with the task node have
Relevant node is as a third relation chain.
In the case of this kind, critical data node is separately as a third relation chain.For example, first by the critical data section
Point is splitted out from second relation chain, as a third relation chain.In remaining node, for critical data section
Each task node in the associated multiple tasks node of point, is traversed using the task node as starting point, will be traversed
All nodes be determined as with the task node have incidence relation node.Assuming that critical data in relation chain shown in Fig. 2 B
Node is that back end 1 tears relation chain shown in Fig. 2 B open as shown in Figure 2 D to be based on back end 1 in the case of this kind
Get the schematic diagram of multiple third relation chains.It should be noted that Fig. 2 B are only shown as example, actual tear open is not represented
Point as a result, for example, should include multiple sections in third relation chain in practical split process other than critical data node
Point, without only including a task node.
The third situation, at least one task section for being directly linked the critical data node and the critical data node
Point and at least one task node there is the node of incidence relation to be split as a third relation chain, it will be except having split
Task node and back end except third relation chain are split as at least one third relation chain.
Wherein, the task node being directly linked with the critical data node refers to the child node as the key business data
Or the task node of father node.In the case of this kind, directly associated at least one task node is split critical data node
For a third relation chain.Wherein, by addition to the third relation chain split task node and back end be split as to
The process of a few third relation chain, and will be in addition to the critical data node and have with the task node in the case of the first
Relevant node as a third relation chain process similarly, this will not be repeated here.Such as, it will again be assumed that shown in Fig. 2 B
Relation chain in critical data node be back end 1, as shown in Figure 2 E in the case of this kind be based on back end 1, to Fig. 2 B
Shown in the schematic diagram of multiple third relation chains that is split of relation chain.
The first situation, can be when migrating third relation chain, if detecting that the relation chain includes critical data section
Then data path mapping table is written in the target store path of the critical data node by point.The second situation and the third situation:
Can key business data first be copied into destination service cluster after the split.
It should be noted that during the second relation chain is split as multiple third relation chains, migration platform can be
Multiple third relation chain adds different relation chain marks.Second relation chain is split as more by above-mentioned three kinds of situations in form
A third relation chain, in order to which the third relation chain that will be obtained by fractionation is distinguished with the first relation chain not split,
Can be that third relation chain addition splits mark, fractionation mark can be embodied in relation chain mark, such as by relation chain
Front two in mark is identified as fractionation.For example, the format of relation chain mark can be xx_yyyy, wherein xx is for indicating
Mark is split, for example 00 indicates the first relation chain not split, 01 indicates the third relation chain obtained by fractionation.Wherein, yyyy
Number for indicating relation chain.
In the present embodiment, during carrying out Data Migration with relation chain, calculating task can still be run, in the mistake of operation
New business data are will produce in journey, due to being limited by network bandwidth, when relation chain is excessive, it is likely that cause to generate new
The speed of business datum is more than the migration velocity of business datum, can cause the relation chain that can not can migrate completion forever in this way, because
High point tethers is split as realizing to indicated by relation chain in the case that small relation chain can ensure normal operation calculating task by this
The migration of data.
Above-mentioned steps 203 and step 204 are that the calculating indicated by the input-output record that is identified according to identical relation chain is appointed
Incidence relation between business and business datum, generates the process of multiple relation chains, and each relation chain includes being used to indicate calculating to appoint
The task node of business, the incidence relation being used to indicate between the back end and task node and back end of business datum.
Above-mentioned steps 202 to 204 are the calculating task daily record according to former service cluster, obtain multiple relation chains, step.Its
In, each relation chain is used to indicate one group of calculating task and business datum with incidence relation.
In the present embodiment, migration platform can be as unit of relation chain, by the business datum indicated by multiple relation chains
It is migrated successively with calculating task to destination service cluster.Wherein, successively migration refer to can be disposably only for a relation chain
Data Migration is carried out, several relation chains can also be directed to and carry out parallel migration.When being migrated based on any one relation chain, just
Often run the calculating task indicated by the relation chain not migrated in multiple relation chains.Wherein, the migration of a relation chain
Journey includes the following steps 205 to 208.
205, for each relation chain in multiple relation chains, multiple business datums indicated by the relation chain are raw
At multiple migration subtasks.
It, can be according to the relation chain meaning during carrying out Data Migration for a relation chain in the present embodiment
The multiple business datums shown, generate multiple migration subtasks, and the process of the multiple migration subtasks of the generation can be:For relationship
Each business datum in multiple business datums indicated by chain executes following procedure:Judge business datum data volume whether
Less than second threshold;If the data volume of business datum is less than second threshold, corresponds to one migration of business datum generation and appoint
Business;It, will according to the time sequencing that data generate according to second threshold if the data volume of business datum is not less than second threshold
Business datum is divided into multiple subservice data, and corresponding each subservice data generate a migration subtask.Wherein, per height
The data volume of business datum is less than second threshold.Wherein, second threshold can be pre-set or changed by migration platform, this
Embodiment is not construed as limiting this.For business datum when storing to service cluster, service cluster can the corresponding record business datum
Storage time, migration platform can determine the generation time of the business datum according to the storage time of record.Wherein, platform is migrated
Can be that configuration information is added in each migration subtask, the configuration information may include corresponding service data former store path and
Target store path.
It should be noted that a business datum shown in the present embodiment refers to the business stored under a store path
Data, when the data volume of business datum is less than second threshold, migration subtask that the corresponding business datum generates just is used for pair
The business datum stored under one store path is migrated.
206, according to multiple migration subtasks, the business datum indicated by the relation chain is moved into destination service cluster.
The business datum can be migrated to target according to the original store path and the target store path and be taken by migration platform
Business cluster.Multiple subtasks corresponding to one relation chain can sequentially execute and can also execute parallel, the present embodiment to this not
It is construed as limiting.
By migrating the business datum indicated by relation chain using different migration subtasks, reduces data and move
The granularity of shifting, and multiple migration subtask can be run parallel, improve the migration of business datum indicated by the relation chain
Efficiency.
In addition, the present embodiment additionally provides the data checking mechanisms to migrating subtask, which can be:
Subtask is migrated for each of multiple migration subtasks, corresponding business datum all moves to mesh in the migration subtask
After marking service cluster, one is carried out to business datum corresponding with the migration subtask in destination service cluster and former service cluster
The verification of cause property;If consistency desired result success, it is determined that the corresponding business datum in migration subtask migrates successfully;If consistent
Property verification failure, it is determined that the migration failure of the corresponding business datum in the migration subtask re-executes the migration subtask.It needs
Illustrate, each configuration information for migrating subtask can also include the data volume size of corresponding service data, should when executing
When migrating subtask, if migration platform, which detects to migrate to the data volume of the business datum of destination service cluster, reaches migration
When data volume size indicated by task, determine that the corresponding business datum in migration subtask has all been migrated to destination service
Cluster.
It should be noted that during being migrated to the business datum indicated by relation chain, the relation chain is signified
The calculating task shown can also continue to run, and therefore, the business datum stored under store path in the relation chain may be sent out
Raw update.For the business datum stored under a store path, the industry that the present embodiment will be stored before production Methods chain
Business data are known as history service data, and newer business datum after production Methods chain is referred to as new business data.It considers
User is less than the possibility to new business data modification to the possibility of history service data modification, therefore, is each moved executing
Move subtask when, can according to sequence of the business datum generated time after arriving first to the business datum under some store path into
Row migration, that is to say and preferentially history service data are migrated, to avoid due to user is modified business datum when need
Business datum is retransmitted, the problem of to reduce transport efficiency.
Wherein, include to the range of the corresponding business datum progress consistency desired result in migration subtask:To business datum
The verification of data volume, the verification for the number of files for including to business datum and the verification of the data content to business datum.Migration
Platform may be used preset algorithm and carry out consistency desired result to the corresponding business datum in the migration subtask, which can be with
It is pre-set, for example, the preset algorithm can be CRC (Cyclic Redundancy Check, cyclic redundancy check code)
Checking algorithm.When data volume of the business datum in former service cluster and destination service cluster, the number of files for including and
When data content is consistent, the consistency desired result success to the business datum is determined.
Wherein, the opportunity for re-executing the migration subtask can be to determine to the mistake of corresponding service consistency verification of data
It is immediately performed, can also be executed after preset time period after determining the failure of corresponding service consistency verification of data after losing, it can be with
The migration subtask of migration failure, the present embodiment pair are re-executed after other corresponding migration subtasks of the relation chain are completed
This is not construed as limiting.
Data check is carried out by being directed to migration subtask, realizes the fine granularity verification of business datum so that work as business
Data Migration malfunction when, can migration subtask granularity on carry out data migration again, compared with the prior art in when
Business datum migrates out the case where needs of staggering the time re-start migration to all business datums, reduces data in transition process and goes out
Wrong cost, improves the efficiency of Data Migration.
In order to reduce influence of the Data Migration to normal use business to the full extent, in the present embodiment, for
The relation chain of migration is not the relevant calculating task out of service during the entire process of migrating the relation chain, but in industry
It is engaged in after Data Migration to certain progress, calculating task is migrated within the calculating task period out of service, with maximum
The calculating task time out of service is reduced in degree.During migrating the business datum indicated by the relation chain, may be used also
To execute following step 206a to step 206d.
Step 206a, it during migrating the business datum indicated by the relation chain, obtains indicated by the relation chain
The migration progress of business datum.
Migrating platform can be according to the total amount of data and the relation chain business datum of the business datum indicated by the relation chain
Data volume of having moved, obtain the migration progress of business datum indicated by the relation chain.For example, business number indicated by the relation chain
According to total amount of data be 100GB, data volume of having moved be 60GB, then can determine the migration of business datum indicated by the relation chain
Progress is 60%.
Step 206b, when the migration progress of business datum is more than default progress, for each of indicated by the relation chain
Calculating task, judges whether the calculating task is held in run-stopping status if the calculating task is in run-stopping status
Row step 206c executes step 206d if the calculating task is in operating status.
Wherein, default progress can be pre-set or changed by migration platform, and the default progress can also be by certainly
Platform is migrated to be adjusted into Mobile state according to network bandwidth, for example, when migration detection of platform is reduced to network bandwidth, it can be appropriate
The numerical value for increasing the default progress, with the time of the reduction migration calculating task institute telephone expenses of maximum possible.
If step 206c, the calculating task is in run-stopping status, maintained before the relation chain completes migration
The run-stopping status of the calculating task.
If step 206d, calculating task is in operating status, wait for calculating task it is out of service after, the relation chain it is complete
At the run-stopping status for maintaining the calculating task before migration.
It should be noted that maintaining the run-stopping status process of calculating task that can claim in step 206c and step 206d
To freeze calculating task process.In order to avoid impacting the business of enterprise customer due to freezing calculating task, freezing to count
It can show the message for freezing calculating task before calculation task to enterprise customer by migration platform, be confirmed by enterprise customer and freezed
And then execute the process for freezing calculating task.
In data migration process, since calculating task is still being run, migrated corresponding to subtask at one
After business datum migration is completed, the business datum is or deleted etc. it is also possible to change, such as change.Cause
This, in order to ensure the integrality of business datum, after the business datum migration indicated by relation chain is completed, migration platform may be used also
As unit of relation chain, to carry out consistency desired result to the business datum indicated by the relation chain, which can be:To target
Business datum in service cluster and former service cluster indicated by the relation chain carries out consistency desired result;If consistency desired result at
Work(then executes subsequent step 207 and step 208;If consistency desired result fails, it is somebody's turn to do according to consistency desired result as a result, determining
The business datum of migration failure indicated by relation chain re-starts migration to the business datum of migration failure.Wherein, to relationship
Can carry out business datum one by one for each migration subtask when business datum indicated by chain carries out consistency desired result
Consistency desired result can also be to carry out the consistency desired result of business datum one by one for each store path in relation chain, and right
In the migration subtask of consistency desired result failure or store path, it is determined that corresponding to the migration subtask or store path
Business datum be migration failure business datum.Migration platform may be used corresponding former migration subtask or re-establish and moves
Subtask is moved, is migrated from newly with the business datum to migration failure, specific transition process is appointed with above-mentioned according to migration
Business carries out the process of Data Migration similarly, and this will not be repeated here.
It should be noted that above-mentioned steps 205 to 206 describe industry indicated by transition relationship chain by taking a relation chain as an example
The process for data of being engaged in.During carrying out Data Migration with relation chain, migration platform can be distinguished according to relation chain mark
The relation chain migrated is the first relation chain without fractionation, or the third relation chain obtained by fractionation.Alternatively, for
A kind of situation that relation chain is split in step 204c includes critical data section due to splitting obtained each third relation chain
Point, migration platform can to the critical data node in each third relation chain add designated identification, by the designated identification come
Identify in the relation chain of migration whether include critical data node, so that it is determined that whether the relation chain of migration is third relation chain.
It should be noted that since the still shared critical data node of the multiple third relation chains obtained by fractionation is signified
The business datum shown, in order to during carrying out Data Migration according to third relation chain, ensure multiple third relationship
Key business data indicated by the shared critical data node of chain synchronizes, the present embodiment use it is double write table mechanism, in data
Two store paths of the key business data are stored in map paths table, one is on the target storage road of destination service cluster
Diameter, the other is in the former store path of former service cluster, which can be:Key business data is obtained in destination service collection
Target store path in group, key business data are the business datum indicated by critical data node;It is mapped in data path
The target store path is added in table, and retains former store path of the key business data in former service cluster.Wherein, to number
It can be executed after relation chain is split according to the process for adding target store path in map paths table, it can also be multiple the
It is executed before the migration of three relation chains, the present embodiment is not construed as limiting this.
During carrying out Data Migration as unit of relation chain, if the relation chain of migration is obtained by fractionation
Third relation chain then writes table mechanism, further comprising the steps of a to step c during being migrated to the third relation chain based on double:
Step a, according to target store path and former store path, in destination service cluster and synchronous pass in former service cluster
Key business datum.
Platform is migrated when detecting that the key business data of former service cluster or destination service cluster updates, root
According to the target store path and former store path, the key business data is carried out in destination service cluster and former service cluster
It is synchronous.
If step b, the business datum indicated by the third relation chain and calculating task have all been migrated to destination service
Cluster is visited then when running the calculating task indicated by third relation chain according to the target store path of data mapping tables record
Ask key business data.
If step c, the business datum and calculating task of third relation chain instruction do not migrate all to destination service collection
Group accesses then when running the calculating task indicated by third relation chain according to the former store path of data map listing record
Key business data.
It, can be according to the mark of service cluster where the data indicated by third relation chain, from data road in the present embodiment
Store path of the acquisition key business data in respective service cluster in diameter mapping table.For example, if indicated by third relation chain
Data in former service cluster, that is to say, the data indicated by the third relation chain are also migrated not successfully to destination service collection
Group when then running the calculating task indicated by the third relation chain, obtains key business data from data path mapping table
Former store path accesses the key business data by the original store path.If the data indicated by third relation chain are in mesh
It marks in service cluster, that is to say, for Successful migration to destination service cluster, then operation should for the data indicated by the third relation chain
When calculating task indicated by third relation chain, the target that key business data is obtained from data path mapping table stores road
Diameter accesses the key business data by the target store path.If Fig. 2 F are to carry out splitting it based on relation chain shown in Fig. 2 B
Afterwards, in the third relation chain obtained to fractionation carries out transition process, the access schematic diagram of key business data, wherein data
Node 1 corresponds to key business data, and 1 place third relation chain of task node is had moved to destination service cluster, task node 2 to
4 place third relation chains do not migrate into destination service cluster also.Calculating task indicated by task node 1 passes through key business
The target store path of data accesses the key business data, the former store path that task node 2 to 4 passes through key business data
Access the key business data.
In conjunction with above-mentioned using double transition process for writing the third relation chain that table mechanism obtains fractionation, table machine is write to double below
The involved flow of system is introduced, and referring to Fig. 2 G, double to write the flow that table mechanism is related to include following processes (1) to (4):
(1) critical data node is obtained.
The process corresponds to the process that critical data node is obtained in the second relation chain.
(2) key business data synchronizes.
Key business data is synchronized in former service cluster and destination service cluster.The process is based on data road
The former store path and target store path of the key business data stored in diameter mapping table.
(3) Intelligent routing of key business data store path.
The position of service cluster residing for third relation chain obtains key business data from data path mapping table
Store path in respective service cluster.Corresponding above-mentioned steps b and step c.
(4) dependence of the third relation chain to the former store path of key business data is gradually released.
After Data Migration to destination service cluster indicated by the third relation chain, the calculating indicated by the third relation chain
Task can that is to say in destination service cluster access critical business datum, relieve the third relation chain and key business
The dependence of the former store path of data.
In the present embodiment, further including moving to the source data of business to the migration of the data indicated by relation chain
It moves, which includes that the data that user inputs in user terminal and user terminal generate be not synchronized to former clothes also in real time
The data of business cluster.In practical applications, which is generally used by calculating task.Specifically, for a relation chain,
The source data can be obtained from real time data processing server by specified interface, and the source data and relation chain are signified
The business datum shown is moved to together in destination service cluster, so as not to influence the normal operation of calculating task.
207, in data path mapping table, original of the business datum indicated by the relation chain in former service cluster is deposited
Storage path is switched to the target store path in destination service cluster.
In the present embodiment, during being migrated to relation chain, migration platform can record relation chain meaning
The target store path for each business datum shown waits for that the business datum indicated by the relation chain is all migrated to destination service collection
When group, for each business datum, migration platform can be in data path mapping table, by the former store path of the business datum
Replace with the target store path of the business datum.
It should be noted that if the business datum is key business data, migration platform is determining and the key business
When business datum corresponding to the relevant all third relation chains of data is migrated to destination service cluster, reflected from the data path
The former store path that the key business data is deleted in firing table retains the target store path of the key business data.
208, the calculating task indicated by relation chain is migrated to destination service cluster.
In the present embodiment, can be by the process that the calculating task indicated by relation chain is migrated to destination service cluster:It obtains
The the first computing resource information and the second computing resource information for taking calculating task, the first computing resource information of calculating task is replaced
It is changed to the second computing resource information.Wherein, the first computing resource information is the meter for being calculating task configuration in former service cluster
Resource information is calculated, the second computing resource information is the computing resource information for being calculating task configuration in destination service cluster.
It should be noted that being migrated to destination service cluster by the calculating task indicated by relation chain, migration is flat
Platform starts all calculating tasks run indicated by the relation chain, to complete the migration to the relation chain.
In addition, the present embodiment during carrying out Data Migration, can also realize that the increment migration of data, the increment are moved
Shifting includes following two levels:
First level migrates the data increased newly in transition process.
During carrying out Data Migration as unit of relation chain, a large amount of calculating tasks in former service cluster still exist
Operation so that former service cluster will produce a large amount of new business data, or former services set after generating multiple relation chains
Calculating task has been increased newly in group, these newly-increased data can be by increasing input-output record body newly in calculating task daily record
It is existing.Platform is migrated to record the calculating according to calculating task daily record, after obtaining multiple relation chains and appoint when generating in daily record
Between input-output record the latest time tag.Migrating platform can be according to the time tag of the record, from former service cluster
The newly-increased input-output record generated after the time tag is obtained in calculating task daily record.
Migration platform can be updated the relation chain not migrated according to the newly-increased input-output record, should
Process can be:Input-output record is increased newly for any bar, if the relation chain not migrated include with this increase newly it is defeated
Enter output and record associated 4th relation chain, then increasing input-output record newly according to this is updated the 4th relation chain;
If in the relation chain not migrated not including the 4th relation chain, input-output record is increased newly according to this and is increased newly with other
Incidence relation between input-output record generates new relation chain, and the process of the generation new relation chain and above-mentioned generation are multiple
Similarly, this will not be repeated here for the process of relation chain.Wherein, increasing associated 4th relation chain of input-output record newly with this is
Refer to, the business datum indicated by the 4th relation chain has with the calculating task indicated by the newly-increased input-output record to be associated with
System, or have with the business datum indicated by the newly-increased input-output record for the calculating task indicated by the 4th relation chain
Incidence relation.
It should be noted that migration platform can be according to the newly-increased input-output record, to the relationship not migrated
The step of chain is updated can execute during relation chain is migrating, can also some relation chain migration complete after
It executes, the present embodiment is not construed as limiting this.The newly-increased input that migration platform can be obtained periodically in calculating task daily record is defeated
Go out record, to be periodically updated to the relation chain not migrated.
In the present embodiment, indicated by transition relationship chain during data, the new calculating generated in former service cluster is appointed
Business may have incidence relation between business datum indicated by the relation chain with the migration, therefore, indicated by the relation chain
After Data Migration to destination service cluster, which then needs from destination service cluster to associated business datum
It is written and read, and since destination service cluster and former service cluster be not in the same IDC computer rooms, the read-write of this kind of business datum
Larger network bandwidth will be occupied, therefore, migrated in first level platform can in time according to calculating task daily record,
The relation chain not migrated is updated so that relation chain can increase newest industry in the former service cluster of comprehensive instruction
Data of being engaged in and calculating task, to avoid the calculating task of former service cluster to read and write the business number of destination service cluster to the full extent
According to the case where, to improve the utilization rate of business processing efficiency and Internet resources.In addition, in order to further avoid former services set
The case where business datum of the calculating task read-write destination service cluster of group, migration platform can also be to owning in former service cluster
The network bandwidth occupancy of calculating task is monitored, and the calculating task of pre-set bandwidths is higher than for network bandwidth occupancy, is moved
Platform is moved preferentially to migrate the relation chain where the calculating task to destination service cluster.
Another level, when migrate interrupt when, based on interrupt when Data Migration state carry out breakpoint transmission.
The process of the breakpoint transmission can be:When being migrated based on any one relation chain, when detecting to relation chain
Migration interrupt operation when, record the migration subtask for not completing migration, stop to the transition process of relation chain;When detecting pair
When the continuation migration operation of relation chain, according to the migration subtask of unfinished migration, by indicated by relation chain business datum and
Calculating task is migrated to destination service cluster.
It should be noted that during carrying out Data Migration with relation chain, it is likely that emergency case can occur and lead
The transition process of the relation chain is caused to interrupt, for example, network failure occurs or has the business datum of higher priority to need immediately
Migration etc..During being migrated to relation chain according to multiple migration subtasks, migration platform can be recorded to multiple
Migration subtask is numbered, and is migrated successively according to the sequence of number.For different migration subtasks, platform is migrated
The state of the migration subtask can be recorded, for example the shape body can be not complete migration, migrating and migrating completion.When moving
When moving migration interrupt operation of the detection of platform to some relation chain, the number for the migration subtask for not completing migration can be recorded.
When detecting the continuation migration operation to the relation chain, the migration subtask for not completing migration is only executed, by the relation chain
The business datum and calculating task not migrated when interruption are migrated to destination service cluster.
In the present embodiment, during carrying out Data Migration with relation chain, migration platform can also use different
Transition state controls transition process, and Fig. 2 H show relation chain in the transition state signal involved in transition process
Figure.Below by taking the transition process of a relation chain as an example, each transition state is introduced:
Start migration:Start to migrate the data indicated by the relation chain.
Obtain source data:After the transition process for determining relation chain starts, the transition state can be entered, moved at this
Under shifting state, migration platform obtains the source data by specified interface from real time data processing server.
User is waited for confirm:When business datum migration progress reaches default progress, before freezing calculating task, to user
It shows that calculating task confirms frozen interface, after being confirmed by user, is transferred to the state for freezing calculating task.It should be noted that
If the relation chain of migration is to split obtained third relation chain, migrates platform and determining the key indicated by third relation chain
When the former store path and target store path of business datum are all contained in data path mapping table, just user is waited for into such
The transition state of confirmation.
Freeze calculating task:Under the state, migration platform execute above-mentioned steps 206c and 206c, all calculating tasks all
When in run-stopping status, into next transition state.
Latency services data are consistent:After freezing calculating task, business datum indicated by relation chain do not migrate completely to
It is in the transition state before destination service cluster.
Service data consistence verifies:The business datum indicated by relation chain is all migrated to destination service cluster, into
Enter the transition state.
Business datum store path switches:After to the success of the consistency desired result of relation chain, into the transition state, execute
To the process of business datum store path switching.
Calculating task migrates:After business datum store path all switches to destination service cluster by former service cluster,
Into the transition state, the process migrated to calculating task is executed.
Defrosting calculating task executes the process for running all calculating tasks, when all meters after calculating task migration is completed
Calculation task all normal operation when, into migration completion status, to complete business datum and calculating task indicated by the relation chain
Migration.
Method provided in this embodiment, by that according to the calculating task daily record in former service cluster, will have incidence relation
Business datum and calculating task indicated using relation chain so that the process of Data Migration is carried out as unit of by relation chain
In, the relation chain migrated will not have an impact other relation chains, the relation chain that still can not be migrated with normal operation
Indicated calculating task, to not interfere with the normal use of business indicated by the relation chain not migrated.
In addition, by carrying out critical data node acquisition to the larger high point tethers of data volume, and by critical data node
Corresponding key business data is arranged to the business datum being able to access that in former service cluster and destination service cluster so that
After high point tethers can be split into multiple small relation chains, no matter small relation chain belongs to former service cluster or destination service collection
Group, can flexibly access the key business data, realize the decoupling between the business of being mutually related, and realize logical
Multiple small relation chains are crossed gradually to be migrated complicated business.
In addition, during being migrated to the data indicated by relation chain, the business indicated by first transition relationship chain
Data, when business datum migration progress reaches default progress, gap that can be out of service in calculating task, to calculating task
It is migrated, greatly reduces influence of the Data Migration to business normal use, and since business datum reaches migration progress
When, remaining business datum amount can usually be completed to migrate in a short period of time, which can be less than the fortune of calculating task
Row period, the process of such Data Migration do not interfere with the normal use of business completely, and the data for realizing user's unaware are moved
It moves.
In addition, by migrating the business datum indicated by relation chain using different migration subtasks, reduce
The granularity of Data Migration, and multiple migration subtask can be run parallel, improve business datum indicated by the relation chain
Transport efficiency.Moreover, when mistake occurs for migration, it is only necessary to migration subtask migrate again, without to whole
The business datum of a service cluster migrates again, reduces the cost of corrupt data in transition process, improves Data Migration
Efficiency.
In addition, by way of breaking the whole up into parts, by former service cluster business datum and calculating task with multiple relationships
Chain is that unit is gradually migrated to destination service cluster, in transition process, the business datum in former service cluster and calculating task
It is constantly reducing so that the hollow remaining server out of former service cluster can be dismantled in resettlement to Target id C computer rooms so that
Server apparatus resource can reuse, and reduce the cost of Data Migration.
Fig. 3 is a kind of block diagram of data migration device provided in an embodiment of the present invention.With reference to Fig. 3, which includes first
Acquiring unit 301 and migration units 302.
Wherein, first acquisition unit 301 is connect with migration units 302, for the calculating task day according to former service cluster
Will obtains multiple relation chains, which is used to record the pass of calculating task and business datum in the original service cluster
Connection relationship, each relation chain are used to indicate one group of calculating task and business datum with incidence relation;Migration units 302 are used
In as unit of relation chain, by indicated by multiple relation chain business datum and calculating task migrated successively to destination service collection
Group;Wherein, when being migrated based on any one relation chain, the relation chain that is not migrated in the multiple relation chain of normal operation
Indicated calculating task.
In one possible implementation, the first acquisition unit is more for being recorded according to the calculating task daily record
Input-output record adds identical relation chain mark, not have association for the input-output record with incidence relation
The input-output record of relationship adds different relation chain marks;Indicated by input-output record according to identical relation chain mark
Calculating task and business datum between incidence relation, generate multiple relation chains, each relation chain includes being used to indicate calculating
The task node of task, the association being used to indicate between the back end and task node and back end of business datum are closed
System.
In one possible implementation, which includes:
Subelement is generated, the calculating task and business indicated by input-output record for being identified according to identical relation chain
Incidence relation between data generates multiple first relation chains;
Subelement is split, if including the second relation chain for multiple first relation chain, by second relation chain
Multiple third relation chains are split as, which is that the data volume of indicated business datum is more than the first pass of first threshold
Tethers.
In one possible implementation, the fractionation subelement is for obtaining multiple back end in second relation chain
Weights, the weights of each back end are used to indicate correlation degree of the back end in second relation chain, and weights are got over
The associated degree of height is higher;According to the position of sequence and multiple back end in second relation chain of weights from high to low
Set, obtain critical data node from multiple back end, the critical data node be in the sequence first can should
Second relation chain is split as the back end of at least two third relation chains;Based on the critical data node, by second relationship
Multiple tasks node associated with the critical data node is split into multiple third relation chains in chain.
In one possible implementation, which is used for:
For each task node in the multiple tasks node that is directly linked with the critical data node, by the key number
Have with the task node when disconnecting relationship according to node, the task node and the critical data node and the task node
The node of incidence relation is determined as third relation chain;Or,
The critical data node is determined as a third relation chain, it is more for being directly linked with the critical data node
In addition to the critical data node and task node is had incidence relation by each service node in a task node
Node be determined as third relation chain;Or,
By the critical data node, with the critical data node be directly linked at least one task node and with this extremely
There is a few task node node of incidence relation to be split as a third relation chain, by the third relation chain in addition to having split
Except task node and back end be split as at least one third relation chain.
In one possible implementation, which is used for for each data in multiple back end
Node, by the data volume of business datum indicated by the number of task node associated with the back end and the back end
Product is determined as the weights of the back end.
In one possible implementation, which further includes:
Second acquisition unit should for obtaining target store path of the key business data in the destination service cluster
Key business data is the business datum indicated by the critical data node;
Adding device for adding the target store path in data path mapping table, and retains the key business number
According to the former store path in the original service cluster.
In one possible implementation, which is used for:
During migrating multiple third relation chain, according to the target store path and the original store path, at this
Destination service cluster and synchronous key business data in the original service cluster;
For any one third relation chain in multiple third relation chain, following procedure is executed:
If the business datum and calculating task indicated by the third relation chain have all been migrated to the destination service cluster,
Then when running the calculating task indicated by the third relation chain, visited according to the target store path of data mapping tables record
Ask the key business data;
If the business datum and calculating task of the third relation chain instruction do not migrate all to the destination service cluster,
Then when running the calculating task indicated by the third relation chain, visited according to the original store path of data map listing record
Ask the key business data.
In one possible implementation, which includes:
Subelement is generated, is used for for each relation chain in multiple relation chain, it is more indicated by the relation chain
A business datum generates multiple migration subtasks, each former store path for migrating subtask and being used to indicate corresponding service data
With target store path;
First migration subelement, for according to multiple migration subtask, the business datum indicated by the relation chain to be moved
Move on to the destination service cluster;
Second migration subelement, for migrating the calculating task indicated by the relation chain to the destination service cluster;
Wherein, when migrating the calculating task indicated by the relation chain, the calculating task indicated by the relation chain, which is in, stops
Only operating status.
In one possible implementation, which is used for:
For each business datum in multiple business datums indicated by the relation chain, following procedure is executed:
Judge whether the data volume of the business datum is less than second threshold;
If the data volume of the business datum is less than the second threshold, corresponds to one migration of business datum generation and appoint
Business;
If the data volume of the business datum is generated according to the second threshold according to data not less than the second threshold
Time sequencing the business datum is divided into multiple subservice data, corresponding each subservice data generate migration and appoint
The data volume of business, each subservice data is less than the second threshold.
In one possible implementation, which is additionally operable to:
During migrating the business datum indicated by the relation chain, the business datum indicated by the relation chain is obtained
Migration progress;
When the migration progress of the business datum is more than default progress, appoint for being calculated each of indicated by the relation chain
Business executes following procedure:
Judge whether the calculating task is in run-stopping status;
If the calculating task is in run-stopping status, the calculating task is maintained before the relation chain completes migration
Run-stopping status;
If the calculating task is in operating status, the calculating task rear, relation chain completion out of service is waited for move
The run-stopping status of the calculating task is maintained before moving.
In one possible implementation, which further includes:
First verification unit is appointed for migrating subtask for each of multiple migration subtask in migration
After corresponding business datum of being engaged in all moves to the destination service cluster, in the destination service cluster and the original service cluster
Business datum corresponding with the migration subtask carries out consistency desired result;If consistency desired result success, it is determined that migration
The corresponding business datum of task migrates successfully;If consistency desired result fails, it is determined that the corresponding business number in the migration subtask
Fail according to migration, re-executes the migration subtask.
In one possible implementation, which further includes:
Second verification unit, for the business indicated by the relation chain in the destination service cluster and the original service cluster
Data carry out consistency desired result;If consistency desired result success, execution move to the calculating task indicated by the relation chain
The step of destination service cluster;If consistency desired result fails, according to consistency desired result as a result, determining the industry of migration failure
Business data re-start migration to the business datum of migration failure.
In one possible implementation, which is used to obtain the first of the calculating task and calculates money
Source information and the second computing resource information, it is calculating task configuration which, which is in the original service cluster,
Computing resource information, which is the calculating money for being calculating task configuration in the destination service cluster
Source information;First computing resource information of the calculating task is replaced with into the second computing resource information.
In one possible implementation, which further includes:
Switch unit, in data path mapping table, original of the business datum in the original service cluster to be stored
Path is switched to the target store path in the destination service cluster.
In one possible implementation, which is additionally operable to when being migrated based on any one relation chain,
When detecting the migration interrupt operation to the relation chain, the migration subtask for not completing migration is recorded, is stopped to the relation chain
Transition process;When detecting the continuation migration operation to the relation chain, the migration subtask of migration is not completed according to this, it will
Business datum and calculating task indicated by the relation chain are migrated to the destination service cluster.
In one possible implementation, which further includes:
Relation chain updating unit, for obtaining newer calculating task daily record;It is right according to the newer calculating task daily record
The relation chain not migrated is updated.
Device provided in this embodiment, by that according to the calculating task daily record in former service cluster, will have incidence relation
Business datum and calculating task indicated using relation chain so that the process of Data Migration is carried out as unit of by relation chain
In, the relation chain migrated will not have an impact other relation chains, the relation chain that still can not be migrated with normal operation
Indicated calculating task, to not interfere with the normal use of business indicated by the relation chain not migrated.
It should be noted that:The data migration device that above-described embodiment provides is in migrating data, only with above-mentioned each function
The division progress of module, can be as needed and by above-mentioned function distribution by different function moulds for example, in practical application
Block is completed, i.e., the internal structure of equipment is divided into different function modules, to complete all or part of work(described above
Energy.In addition, the data migration device that above-described embodiment provides belongs to same design with data migration method embodiment, it is specific real
Existing process refers to embodiment of the method, and which is not described herein again.
Fig. 4 is a kind of block diagram of data migration device provided in an embodiment of the present invention.For example, device 400 can be provided
For a server.With reference to Fig. 4, device 400 includes processing component 422, further comprises one or more processors, Yi Jiyou
Memory resource representated by memory 432, can be by the instruction of the execution of processing component 422, such as application program for storing.
The application program stored in memory 432 may include it is one or more each correspond to one group of instruction module.
In addition, processing component 422 is configured as executing instruction, to execute in above-mentioned data migration method embodiment performed by server
Method.
Device 400 can also include the power management that a power supply module 426 is configured as executive device 400, and one has
Line or radio network interface 450 are configured as device 400 being connected to network and input and output (I/O) interface 458.Dress
Setting 400 can operate based on the operating system for being stored in memory 432, such as Windows ServerTM, Mac OS XTM,
UnixTM,LinuxTM, FreeBSDTMOr it is similar.
In the exemplary embodiment, it includes the non-transitorycomputer readable storage medium instructed, example to additionally provide a kind of
Such as include the memory of instruction, above-metioned instruction can be executed by the processor in server and be moved with the data completed in above-described embodiment
Shifting method.For example, the non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-
ROM, tape, floppy disk and optical data storage devices etc..
One of ordinary skill in the art will appreciate that realizing that all or part of step of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can be stored in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (15)
1. a kind of data migration method, which is characterized in that the method includes:
According to the calculating task daily record of former service cluster, multiple relation chains are obtained, the calculating task daily record is described for recording
The incidence relation of calculating task and business datum in former service cluster, each relation chain are used to indicate one group with incidence relation
Calculating task and business datum;
As unit of relation chain, by indicated by the multiple relation chain business datum and calculating task migrate successively to target take
Business cluster;
Wherein, when being migrated based on any one relation chain, the pass that is not migrated in the multiple relation chain of normal operation
Calculating task indicated by tethers.
2. according to the method described in claim 1, it is characterized in that, the calculating task daily record according to former service cluster, is obtained
Multiple relation chains are taken to include:
According to a plurality of input-output record that the calculating task daily record is recorded, for the input-output record with incidence relation
Identical relation chain mark is added, different relation chain marks is added for the input-output record without incidence relation;
The incidence relation between the calculating task and business datum indicated by input-output record according to identical relation chain mark,
Multiple relation chains are generated, each relation chain includes the task node for being used to indicate calculating task, the number for being used to indicate business datum
According to the incidence relation between node and task node and back end.
3. according to the method described in claim 2, it is characterized in that, the input-output record identified according to identical relation chain
Incidence relation between indicated calculating task and business datum, generating multiple relation chains includes:
The incidence relation between the calculating task and business datum indicated by input-output record according to identical relation chain mark,
Generate multiple first relation chains;
If the multiple first relation chain includes the second relation chain, second relation chain is split as multiple thirds and is closed
Tethers, second relation chain are that the data volume of indicated business datum is more than the first relation chain of first threshold.
4. according to the method described in claim 3, it is characterized in that, described be split as multiple thirds passes by second relation chain
Tethers includes:
The weights of multiple back end in second relation chain are obtained, the weights of each back end are used to indicate the data
Correlation degree of the node in second relation chain, the more high associated degree of weights are higher;
According to the position of sequence and the multiple back end in second relation chain of weights from high to low, from described more
Critical data node is obtained in a back end, the critical data node can be by described second for first in the sequence
Relation chain is split as the back end of at least two third relation chains;
Based on the critical data node, by multiple tasks associated with the critical data node in second relation chain
Node is split into multiple third relation chains.
5. according to the method described in claim 4, it is characterized in that, described be based on the critical data node, by described second
Multiple tasks node associated with the critical data node, which is split to multiple third relation chains, in relation chain includes:
For each task node in the multiple tasks node that is directly linked with the critical data node, by the crucial number
When disconnecting relationship according to node, the task node and the critical data node and the task node with the task
There is node the node of incidence relation to be determined as third relation chain;Or,
The critical data node is determined as a third relation chain, it is more for being directly linked with the critical data node
Each service node in a task node, by it is in addition to the critical data node, with the task node have be associated with
The node of relationship is determined as third relation chain;Or,
By the critical data node, with the critical data node be directly linked at least one task node and with it is described
There is at least one task node the node of incidence relation to be split as a third relation chain, by the third relationship in addition to having split
Task node and back end except chain are split as at least one third relation chain.
6. according to the method described in claim 4, it is characterized in that, described be split as multiple thirds passes by second relation chain
After tethers, the method further includes:
Target store path of the key business data in the destination service cluster is obtained, the key business data is described
Business datum indicated by critical data node;
The target store path is added in data path mapping table, and retains the key business data in the former service
Former store path in cluster.
7. according to the method described in claim 1, it is characterized in that, described as unit of relation chain, by the multiple relation chain
Indicated business datum and calculating task, which is migrated successively to destination service cluster, includes:
For each relation chain in the multiple relation chain, multiple business datums indicated by the relation chain generate
Multiple migration subtasks, each former store path and target store path for migrating subtask and being used to indicate corresponding service data;
According to the multiple migration subtask, the business datum indicated by the relation chain is moved into the destination service collection
Group;
Calculating task indicated by the relation chain is migrated to the destination service cluster;
Wherein, when migrating the calculating task indicated by the relation chain, the calculating task indicated by the relation chain, which is in, stops
Only operating status.
8. the method according to the description of claim 7 is characterized in that the business datum by indicated by the relation chain migrates
Include to the destination service cluster:
During migrating the business datum indicated by the relation chain, the business datum indicated by the relation chain is obtained
Migration progress;
When the migration progress of the business datum is more than default progress, appoint for being calculated each of indicated by the relation chain
Business executes following procedure:
Judge whether the calculating task is in run-stopping status;
If the calculating task is in run-stopping status, described calculate is maintained to appoint before the relation chain completes migration
The run-stopping status of business;
If the calculating task is in operating status, the calculating task rear, described relation chain out of service is waited for complete
The run-stopping status of the calculating task is maintained before migration.
9. the method according to the description of claim 7 is characterized in that the calculating task by indicated by the relation chain migrates
Include to the destination service cluster:
Obtain the first computing resource information and the second computing resource information of the calculating task, the first computing resource information
To be the computing resource information of the calculating task configuration in the former service cluster, the second computing resource information be
It is the computing resource information of the calculating task configuration in the destination service cluster;
First computing resource information of the calculating task is replaced with into the second computing resource information.
10. the method according to the description of claim 7 is characterized in that the business datum by indicated by the relation chain is moved
It moves on to after the destination service cluster, the method further includes:
In data path mapping table, former store path of the business datum in the former service cluster is switched in institute
State the target store path in destination service cluster.
11. a kind of data migration device, which is characterized in that described device includes:
First acquisition unit obtains multiple relation chains, the calculating task for the calculating task daily record according to former service cluster
Daily record is used to record the incidence relation of calculating task and business datum in the former service cluster, and each relation chain is used to indicate tool
One group of relevant calculating task and business datum;
Migration units, for as unit of relation chain, by indicated by the multiple relation chain business datum and calculating task according to
It is secondary to migrate to destination service cluster;
Wherein, when being migrated based on any one relation chain, the pass that is not migrated in the multiple relation chain of normal operation
Calculating task indicated by tethers.
12. according to the devices described in claim 11, which is characterized in that the first acquisition unit is used to be appointed according to described calculate
The a plurality of input-output record that business daily record is recorded adds identical relation chain mark for the input-output record with incidence relation
Know, different relation chain marks is added for the input-output record without incidence relation;According to the defeated of identical relation chain mark
Enter the incidence relation between output record indicated calculating task and business datum, generates multiple relation chains, each relation chain
Including being used to indicate the task node of calculating task, the back end for being used to indicate business datum and task node and data section
Incidence relation between point.
13. device according to claim 12, which is characterized in that the first acquisition unit includes:
Subelement is generated, the calculating task and business datum indicated by input-output record for being identified according to identical relation chain
Between incidence relation, generate multiple first relation chains;
Subelement is split, if including the second relation chain for the multiple first relation chain, by second relation chain
Multiple third relation chains are split as, second relation chain is that the data volume of indicated business datum is more than the first of first threshold
Relation chain.
14. device according to claim 13, which is characterized in that the fractionation subelement is for obtaining second relationship
The weights of the weights of multiple back end in chain, each back end are used to indicate the back end in second relation chain
In correlation degree, the more high associated degree of weights is higher;According to weights sequence from high to low and the multiple back end
Position in second relation chain obtains critical data node, the critical data section from the multiple back end
Point is first back end that second relation chain can be split as at least two third relation chains in the sequence;Base
In the critical data node, multiple tasks node associated with the critical data node in second relation chain is torn open
Divide into multiple third relation chains.
15. device according to claim 14, which is characterized in that the fractionation subelement is used for:
For each task node in the multiple tasks node that is directly linked with the critical data node, by the crucial number
When disconnecting relationship according to node, the task node and the critical data node and the task node with the task
There is node the node of incidence relation to be determined as third relation chain;Or,
The critical data node is determined as a third relation chain, it is more for being directly linked with the critical data node
Each service node in a task node, by it is in addition to the critical data node, with the task node have be associated with
The node of relationship is determined as third relation chain;Or,
By the critical data node, with the critical data node be directly linked at least one task node and with it is described
There is at least one task node the node of incidence relation to be split as a third relation chain, by the third relationship in addition to having split
Task node and back end except chain are split as at least one third relation chain.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710197702.7A CN108664496B (en) | 2017-03-29 | 2017-03-29 | Data migration method and device |
PCT/CN2018/078398 WO2018177107A1 (en) | 2017-03-29 | 2018-03-08 | Data migration method, migration server, and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710197702.7A CN108664496B (en) | 2017-03-29 | 2017-03-29 | Data migration method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108664496A true CN108664496A (en) | 2018-10-16 |
CN108664496B CN108664496B (en) | 2022-03-25 |
Family
ID=63674187
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710197702.7A Active CN108664496B (en) | 2017-03-29 | 2017-03-29 | Data migration method and device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN108664496B (en) |
WO (1) | WO2018177107A1 (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110490322A (en) * | 2019-08-14 | 2019-11-22 | 北京中科寒武纪科技有限公司 | Method for splitting and device, the electronic equipment and storage medium of operation node |
CN110503199A (en) * | 2019-08-14 | 2019-11-26 | 北京中科寒武纪科技有限公司 | Method for splitting and device, the electronic equipment and storage medium of operation node |
CN110597609A (en) * | 2019-09-17 | 2019-12-20 | 深圳市及响科技有限公司 | Cluster migration and automatic recovery method and system |
CN111274230A (en) * | 2020-03-26 | 2020-06-12 | 北京奇艺世纪科技有限公司 | Data migration management method, device, equipment and storage medium |
CN111459411A (en) * | 2020-03-30 | 2020-07-28 | 北京奇艺世纪科技有限公司 | Data migration method, device, equipment and storage medium |
CN113438267A (en) * | 2020-03-23 | 2021-09-24 | 华为技术有限公司 | Method and equipment for analyzing stream data |
CN116049096A (en) * | 2022-05-05 | 2023-05-02 | 荣耀终端有限公司 | Data migration method, electronic equipment and storage medium |
Families Citing this family (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110399356B (en) * | 2019-06-14 | 2023-02-24 | 阿里巴巴集团控股有限公司 | Online data migration method and device, computing equipment and storage medium |
CN110989929A (en) * | 2019-11-22 | 2020-04-10 | 浪潮电子信息产业股份有限公司 | MON service migration method, device, equipment and readable storage medium |
CN113051245A (en) * | 2019-12-26 | 2021-06-29 | 云丁网络技术(北京)有限公司 | Method, device and system for migrating data |
CN111258985A (en) * | 2020-01-17 | 2020-06-09 | 中国工商银行股份有限公司 | Data cluster migration method and device |
CN111708755A (en) * | 2020-05-20 | 2020-09-25 | 北京奇艺世纪科技有限公司 | Data migration method, device, system, electronic equipment and readable storage medium |
CN111708763B (en) * | 2020-06-18 | 2023-12-01 | 北京金山云网络技术有限公司 | Data migration method and device of sliced cluster and sliced cluster system |
CN114024956B (en) * | 2020-07-17 | 2024-03-12 | 北京达佳互联信息技术有限公司 | Data migration method, device, server and storage medium |
CN112506606A (en) * | 2020-11-23 | 2021-03-16 | 北京达佳互联信息技术有限公司 | Migration method, device, equipment and medium for containers in cluster |
CN112653539B (en) * | 2020-12-29 | 2023-06-20 | 杭州趣链科技有限公司 | Storage method, device and equipment for data to be stored |
CN113535087B (en) * | 2021-07-13 | 2023-10-17 | 咪咕互动娱乐有限公司 | Data processing method, server and storage system in data migration process |
CN116954870B (en) * | 2023-09-19 | 2024-02-02 | 苏州元脑智能科技有限公司 | Migration method, recovery method and device of cross-cluster application and cluster system |
Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6047323A (en) * | 1995-10-19 | 2000-04-04 | Hewlett-Packard Company | Creation and migration of distributed streams in clusters of networked computers |
US20040153451A1 (en) * | 2002-11-15 | 2004-08-05 | John Phillips | Methods and systems for sharing data |
US20070118710A1 (en) * | 2005-11-18 | 2007-05-24 | Hiroshi Yamakawa | Storage system and data migration method |
CN102855299A (en) * | 2012-08-16 | 2013-01-02 | 上海引跑信息科技有限公司 | Method for realizing iterative migration of distributed database without interrupting service |
CN103164261A (en) * | 2011-12-15 | 2013-06-19 | 中国移动通信集团公司 | Multicenter data task processing method, multicenter data task processing device and multicenter data task processing system |
CN103647849A (en) * | 2013-12-24 | 2014-03-19 | 华为技术有限公司 | Method and device for migrating businesses and disaster recovery system |
CN103955491A (en) * | 2014-04-15 | 2014-07-30 | 南威软件股份有限公司 | Method for synchronizing timing data increment |
CN103970879A (en) * | 2014-05-16 | 2014-08-06 | 中国人民解放军国防科学技术大学 | Method and system for regulating storage positions of data blocks |
CN104184813A (en) * | 2014-08-20 | 2014-12-03 | 杭州华为数字技术有限公司 | Load balancing method of virtual machines, related equipment and trunking system |
CN104935618A (en) * | 2014-03-19 | 2015-09-23 | 福建福昕软件开发股份有限公司北京分公司 | Cluster disposition method |
CN105404474A (en) * | 2015-12-07 | 2016-03-16 | 上海爱数信息技术股份有限公司 | Data migration method of heterogeneous distributed memory system |
CN106055670A (en) * | 2016-06-06 | 2016-10-26 | 中国工商银行股份有限公司 | Inter-system data migration method and device |
CN106202212A (en) * | 2016-06-28 | 2016-12-07 | 微梦创科网络科技(中国)有限公司 | A kind of method and system realizing data fractionation based on data server cluster |
CN102999537B (en) * | 2011-09-19 | 2017-01-18 | 阿里巴巴集团控股有限公司 | System and method for data migration |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102982085B (en) * | 2012-10-31 | 2017-05-31 | 北京奇虎科技有限公司 | Data mover system and method |
CN103324466B (en) * | 2013-05-24 | 2017-05-03 | 浪潮电子信息产业股份有限公司 | Data dependency serialization IO parallel processing method |
-
2017
- 2017-03-29 CN CN201710197702.7A patent/CN108664496B/en active Active
-
2018
- 2018-03-08 WO PCT/CN2018/078398 patent/WO2018177107A1/en active Application Filing
Patent Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6047323A (en) * | 1995-10-19 | 2000-04-04 | Hewlett-Packard Company | Creation and migration of distributed streams in clusters of networked computers |
US20040153451A1 (en) * | 2002-11-15 | 2004-08-05 | John Phillips | Methods and systems for sharing data |
US20070118710A1 (en) * | 2005-11-18 | 2007-05-24 | Hiroshi Yamakawa | Storage system and data migration method |
CN102999537B (en) * | 2011-09-19 | 2017-01-18 | 阿里巴巴集团控股有限公司 | System and method for data migration |
CN103164261A (en) * | 2011-12-15 | 2013-06-19 | 中国移动通信集团公司 | Multicenter data task processing method, multicenter data task processing device and multicenter data task processing system |
CN102855299A (en) * | 2012-08-16 | 2013-01-02 | 上海引跑信息科技有限公司 | Method for realizing iterative migration of distributed database without interrupting service |
CN103647849A (en) * | 2013-12-24 | 2014-03-19 | 华为技术有限公司 | Method and device for migrating businesses and disaster recovery system |
CN104935618A (en) * | 2014-03-19 | 2015-09-23 | 福建福昕软件开发股份有限公司北京分公司 | Cluster disposition method |
CN103955491A (en) * | 2014-04-15 | 2014-07-30 | 南威软件股份有限公司 | Method for synchronizing timing data increment |
CN103970879A (en) * | 2014-05-16 | 2014-08-06 | 中国人民解放军国防科学技术大学 | Method and system for regulating storage positions of data blocks |
CN104184813A (en) * | 2014-08-20 | 2014-12-03 | 杭州华为数字技术有限公司 | Load balancing method of virtual machines, related equipment and trunking system |
CN105404474A (en) * | 2015-12-07 | 2016-03-16 | 上海爱数信息技术股份有限公司 | Data migration method of heterogeneous distributed memory system |
CN106055670A (en) * | 2016-06-06 | 2016-10-26 | 中国工商银行股份有限公司 | Inter-system data migration method and device |
CN106202212A (en) * | 2016-06-28 | 2016-12-07 | 微梦创科网络科技(中国)有限公司 | A kind of method and system realizing data fractionation based on data server cluster |
Non-Patent Citations (3)
Title |
---|
ANJIA STRUNK: "Costs of virtual machine live migration A survey", 《IEEE EIGHTH WORLD CONGRESS ON SERVICES》 * |
姜游: "分组技术及其在集群中的应用", 《计算机系统应用》 * |
龚卫华: "数据库集群系统的关键技术研究", 《中国博士学位论文全文数据库_信息科技辑》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110490322A (en) * | 2019-08-14 | 2019-11-22 | 北京中科寒武纪科技有限公司 | Method for splitting and device, the electronic equipment and storage medium of operation node |
CN110503199A (en) * | 2019-08-14 | 2019-11-26 | 北京中科寒武纪科技有限公司 | Method for splitting and device, the electronic equipment and storage medium of operation node |
CN110597609A (en) * | 2019-09-17 | 2019-12-20 | 深圳市及响科技有限公司 | Cluster migration and automatic recovery method and system |
CN113438267A (en) * | 2020-03-23 | 2021-09-24 | 华为技术有限公司 | Method and equipment for analyzing stream data |
CN113438267B (en) * | 2020-03-23 | 2023-02-28 | 华为技术有限公司 | Method and equipment for analyzing stream data |
CN111274230A (en) * | 2020-03-26 | 2020-06-12 | 北京奇艺世纪科技有限公司 | Data migration management method, device, equipment and storage medium |
CN111274230B (en) * | 2020-03-26 | 2024-03-08 | 北京奇艺世纪科技有限公司 | Data migration management method, device, equipment and storage medium |
CN111459411A (en) * | 2020-03-30 | 2020-07-28 | 北京奇艺世纪科技有限公司 | Data migration method, device, equipment and storage medium |
CN111459411B (en) * | 2020-03-30 | 2023-07-21 | 北京奇艺世纪科技有限公司 | Data migration method, device, equipment and storage medium |
CN116049096A (en) * | 2022-05-05 | 2023-05-02 | 荣耀终端有限公司 | Data migration method, electronic equipment and storage medium |
CN116049096B (en) * | 2022-05-05 | 2024-04-16 | 荣耀终端有限公司 | Data migration method, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108664496B (en) | 2022-03-25 |
WO2018177107A1 (en) | 2018-10-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108664496A (en) | Data migration method and device | |
US11086555B1 (en) | Synchronously replicating datasets | |
CN109729129B (en) | Configuration modification method of storage cluster system, storage cluster and computer system | |
WO2019154394A1 (en) | Distributed database cluster system, data synchronization method and storage medium | |
CN106528327B (en) | A kind of data processing method and backup server | |
EP2784675B1 (en) | Method, device and system for data reconstruction | |
US20230004299A1 (en) | Continuing Replication During Storage System Transportation | |
US8862541B1 (en) | N-site asynchronous replication | |
KR102051282B1 (en) | Network-bound memory with optional resource movement | |
CN108628717A (en) | A kind of Database Systems and monitoring method | |
US20180004777A1 (en) | Data distribution across nodes of a distributed database base system | |
CN102597958A (en) | Symmetric live migration of virtual machines | |
EP3015998A1 (en) | Zoning balance subtask delivering method, apparatus and system | |
US11599554B2 (en) | Synchronizing changes to stale components of a distributed object using synchronization bitmaps | |
WO2008085483A2 (en) | Toggling between concurrent and cascaded triangular asynchronous replication | |
US10445295B1 (en) | Task-based framework for synchronization of event handling between nodes in an active/active data storage system | |
JP2021135703A (en) | Information processing system and method | |
WO2019199419A1 (en) | High throughput order fullfillment database system | |
CN111680019B (en) | Block chain data expansion method and device | |
CN110377664B (en) | Data synchronization method, device, server and storage medium | |
CN105740049A (en) | Control method and apparatus | |
CN105635285B (en) | A kind of VM migration scheduling method based on state aware | |
US20230281167A1 (en) | Hybrid synchronization using a shadow component | |
CN109254873A (en) | Data back up method, relevant apparatus and system | |
CN105488139B (en) | The method of cross-platform storing data migration based on power information acquisition system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |