CN103106286B - Method and device for managing metadata - Google Patents
Method and device for managing metadata Download PDFInfo
- Publication number
- CN103106286B CN103106286B CN201310068332.9A CN201310068332A CN103106286B CN 103106286 B CN103106286 B CN 103106286B CN 201310068332 A CN201310068332 A CN 201310068332A CN 103106286 B CN103106286 B CN 103106286B
- Authority
- CN
- China
- Prior art keywords
- node
- metadata
- user
- updated
- preserving
- 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.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method and a device for managing metadata. The method comprises the following steps: allocating the metadata into a plurality of assemblages; saving the plurality of assemblages of the metadata in a mode of full copy at each node of a plurality of nodes; and according to a request of a user, accessing partial or total assemblages which are saved in the nodes. According to the invention, the metadata are allocated into the plurality of assemblages and are saved in the nodes and then service is provided for the user, so that the resource of the metadata can be integrated and the storage structure of the metadata is optimized. Therefore, the efficiency of updating the nodes of the metadata is increased. In addition, as the mode of full copy is adopted, the data loss is reduced and the resource utilization rate of distributed data storage systems is increased.
Description
Technical field
The present invention relates to computer realm, and especially, it is related to a kind of management method of metadata and its device.
Background technology
It is important step in distributed memory system that metadata is carried out storing, in recent years, with data volume drastically
Increase, distributed storage technology becomes commonly used technology, and various distributed file systems, distributed data base software layer go out
Not poor, these technology employ different mechanism and metadata are stored.In actual applications, read operation is carried out to metadata
It is very frequently, in order to ensure the high efficiency of distributed system, specific data structure storage can be become to exist metadata organization
In internal memory.Metadata node is the key modules affecting whole distributed memory system availability.User side is to distributed system
Read-write operation will be through metadata node.Metadata node is typically necessary has larger internal memory, and because in first number
Externally provide it may appear that multi-user end concurrently accesses the situation of same metadata node during service according to node, therefore, it needs again
There is larger bandwidth.
At present, most distributed memory system adopts single-point to provide metadata clothes to reduce the burden of metadata management
The mode of business, for the sake of security, and with the addition of metadata backup node in systems, is entering row write to metadata node every time
During operation, need to write two parts of metadata (that is, simultaneously carrying out write operation to metadata node and backup node).When cell data section
During point failure, metadata backup node in addition will continue to provide service, prevents distributed memory system from paralysing.Such as, back up
Node, after service node fault, is used for being recovered, but the time of this mode fault recovery is very long.For example, distributed
File system (hadoop distributed file system, abbreviation hdfs) and Google's file system (google file
System, abbreviation gfs) earlier version, all using this scheme, but hdfs typically require half an hour could recover therefore
Barrier, and hot-swap cannot be realized, lead to cluster externally to provide the ability of service to substantially reduce.
Therefore, entirely replicate storage externally providing service using meta-data distribution formula will be trend of the times simultaneously.Distributed number
According to storehouse (hadoop database, abbreviation hbase) using by the way of although there being the mechanism of copy and hot-swap, but still
So only one of which copy externally provides service, and also can produce bottleneck problem in high concurrent.If using many copies simultaneously
Externally provide service, then easily lead to the inconsistent of reading and writing data.Here it is what availability and concordance cannot meet simultaneously
Classical theory.And to metadata when being updated operation, if using conventional two benches in distributed system or three ranks
The data that section is submitted to, thus being updated the method operating, can substantially reduce availability of data.
For not high to the metadata updates operating efficiency in distributed data-storage system in correlation technique, lead to system
The problem that availability of data reduces, not yet proposes effective solution at present.
Content of the invention
For not high to the metadata updates operating efficiency in distributed data-storage system in correlation technique, lead to system
The problem that availability of data reduces, the present invention proposes a kind of management method of metadata, can optimize the storage organization of metadata,
Thus improving the efficiency that metadata node is updated, provide a kind of new think of for realizing metadata management easily and effectively
Road, and then improve the resource utilization of distributed data-storage system.
The technical scheme is that and be achieved in that:
According to an aspect of the invention, it is provided a kind of management method of metadata.
This management method includes:
Metadata configurations are multiple set;
Multiple set in the Preservation Metadata in the way of full duplication of each node in multiple nodes;
According to the part or all of set preserving in the request accessed node of user.
And, this management method further includes:
Configure each set of metadata by the form of form, each form includes the title of metadata in this set
With corresponding content.
Alternatively, the above includes at least one of:
The type of metadata, the descriptive information of metadata.
Additionally, in the case that the part or all of set preserving in node is updated, using affair mechanism, Yi Ji
It is combined into unit to be updated.
Further, in the case that the part or all of set preserving in a node is updated, if collected to one
Conjunction is updated successfully, and there is, with this set, the set that associate and updates unsuccessfully, then associate to this set and with improvement and presence
Set execution rollback.
In addition, this management method further includes:
In the case that node is updated, according to the predetermined sequence of multiple nodes, the node that needs are updated is successively
Carry out unidirectional renewal.
And, this management method further includes:
During node is updated, if updating the number of nodes being successfully updated when stopping to be less than predetermined threshold,
Execute rollback to updating the node completing.
Further, in the case of determining that according to the request of user user needs to read metadata, asking according to user
The part or all of set preserving in accessed node is asked to include:
Determine the version of the metadata that user needs to read, the metadata preserving according to multiple nodes according to the request of user
Version and multiple node load, by one of multiple nodes node by need read metadata be supplied to user.
According to another aspect of the present invention, there is provided a kind of managing device of metadata.
This managing device includes:
Configuration module, for being multiple set by metadata configurations;
Preserving module, for the multiple collection in the Preservation Metadata in the way of full duplication of each node in multiple nodes
Close;
Access modules, for the part or all of set preserving in the request accessed node according to user.
Wherein, access modules are further used in the case that node is updated, according to the predetermined row of multiple nodes
Sequence, carries out unidirectional renewal successively to the node needing to update.
The present invention is saved in node by the form being multiple set by metadata configurations, then provides service to user,
The resource of metadata can be integrated, optimize the storage organization of metadata, thus improving the efficiency that metadata node is updated,
And due to employing the mode of full duplication, reduce the probability of loss of data, and then improve distributed data-storage system
Resource utilization.
Brief description
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to institute in embodiment
Need use accompanying drawing be briefly described it should be apparent that, drawings in the following description be only the present invention some enforcement
Example, for those of ordinary skill in the art, on the premise of not paying creative work, can also obtain according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 is the flow chart of the management method of metadata according to embodiments of the present invention;
Fig. 2 is the schematic diagram of the model that the embodiment of the present invention adopts relational data library management Cluster Database;
Fig. 3 is that metadata management method according to embodiments of the present invention enters when minsuc (minimum successful node number) is 1
The schematic diagram of row metadata management;
Fig. 4 is that metadata management method according to embodiments of the present invention is carried out showing of metadata management when minsuc is 2
It is intended to;
Fig. 5 is the block diagram of meta data management device according to an embodiment of the invention.
Specific implementation method
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation description is it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, the every other embodiment that those of ordinary skill in the art are obtained, broadly fall into present invention protection
Scope.
According to embodiments of the invention, there is provided a kind of management method of metadata.
As shown in figure 1, the management method of metadata according to embodiments of the present invention includes:
Step s101, metadata configurations are multiple set;
Step s103, in multiple set of the Preservation Metadata in the way of full duplication of each node in multiple nodes;
Step s105, according to the part or all of set preserving in the request accessed node of user.
Management method according to embodiments of the present invention each of configuration metadata can be collected in the form of passing through form further
Close.In actual applications, the microcosmic storage that metadata is properly termed as metadata management again is configured by the form of form.Micro- at this
See in storage, metadata is designed as form storage, and the up-to-date modification of every data is log forever, will more according to configuration
New operation is persisted on one or more node.
Wherein, the corresponding form of each set includes the title of each metadata and corresponding content in this set.
Alternatively, the content in set can further include at least one of: the type of metadata, the explanation letter of metadata
Breath.
Table 1 take partition information as the storage organization being illustrated metadata table according to embodiments of the present invention:
Field name | Type | Explanation |
dist_id | number | Partition number |
table_name | varchar(10) | Data table name |
part_column | varchar(10) | Subregion row name |
dist_rule | number | Zoning ordinance |
node_num | number | Interstitial content |
Table 1
As shown in table 1, metadata can include field name (i.e. the title of metadata), type and explanation.Different
Field name distinguishes the metadata representing different, dist_id, table_name, the part_ representing including but not limited in table 1
column、dist_rule、node_num.The type of metadata can include but is not limited in table 1 represent number,
varchar(10).Explanation can include but is not limited to the partition number of expression in table 1, data table name, subregion row name, subregion rule
Then, interstitial content.
Metadata table shown in table 1, for the zoning ordinance of form, with form for unit organization metadata, and multiple
Miscellaneous structure sequence mode stores compares, it is to avoid adds the complicated work such as index, can make full use of existing relationship type number
According to the data organizing tool in storehouse, data access is allowed to be easier to position.
It should be noted that being only a kind of storage organization of exemplary metadata table, the storage organization of metadata table shown in table 1
Scheme be not limited to the situation shown in table, in practice can with using other mode configuration forms to represent unit
The content of data.
Additionally, the management method of metadata according to embodiments of the present invention, the part or all of collection preserving in node
In the case that conjunction is updated, it is possible to use affair mechanism, by gather in units of be updated.
In addition, in one embodiment, when in a node, the part or all of set of preservation is updated, if right
One set is updated successfully, and there is, with this set, the set associating and update unsuccessfully, then close to this set and with improving and existing
The set execution rollback of connection.
Fig. 2 shows a kind of application mode of metadata management method according to embodiments of the present invention, i.e. adopt relationship type
Data base (relative database, abbreviation rdb) manages metadata.Taking single node as a example, can be by the unit of this cell node
Data is divided into multiple memory element (corresponding to above-mentioned set, or form), (complete by the characteristic of relevant database
Property, functional dependence etc.) and in view of relatedness between metadata, it is possible to achieve each memory element (that is, set or table
Lattice) between association.
In fig. 2, in single node, the metadata of storage can include table 1 to table 6, wherein, each table corresponding information respectively
For table 1 mainly stores form data, table name, row name, Distinguish;Table 2 mainly stores the distributed intelligence of form;Table 3 is mainly deposited
Storage data hash partition information;Main data storage scope (range) partition information of table 4;The main data storage poll subregion of table 5
Information;The main data storage nodal information of table 6.
Form in Fig. 2 can be modified, for example, it is possible to modify to partitioned mode.In modification partitioned mode
When, after table 1 is modified, by way of external key it is necessary to modification, add or delete table 3, table 4 and/or table 5 content (i.e.
There is the set associating corresponding to above-mentioned with this set).For example, when needing to be changed to certain form divisional type by poll
Hash, then only need to the information in deletion table 5, increases information in table 3.If the partly success of operation, part is lost
Lose, direct rollback can (correspond to mentioned above: the situation that the part or all of set preserving in a node is updated
Under, if be updated successfully to a set, and there is, with this set, the set that associates and update unsuccessfully, then to this set and with this collection
Close the set execution rollback that there is association).
By above-mentioned operation management mode, on the one hand, metadata information can be divided into multiple the patrolling of presence contact
Collect unit (database table or set), thus effectively being managed.On the other hand, metadata letter in renewal single node
Concordance and the integrity of metadata information during breath, can be realized by the affair mechanism of relational data and storing process.
Each set adopting relational data library management metadata above is only specific example, in practical application
In, can be set up on each node by means of the other modes outside relevant database completely and manage metadata each
Set (form).
In the case that user side accesses metadata, the difference of the metadata type that can access as needed, only access
Certain form, when multiple user sides access different forms, operation does not interfere with each other.If necessary to be updated to form, then only
The form that oneself is accessed is needed to lock.
This metadata WriteMode reduces the granularity of locking, improves the efficiency of concurrent write metadata.
When the metadata of Cluster Database is managed, can be multiple by metadata information in the way of using full duplication
Make and stored on multiple nodes, thus increasing the reliability of metadata information in single node.
Metadata management method according to embodiments of the present invention, the synchronization between following descriptive metadata node and recovery.
During node is being updated, if updating the number of nodes being successfully updated when stopping to be less than predetermined threshold, to having updated
The node execution rollback becoming.Further, in the case of determining that according to the request of user user needs to read metadata, according to
The part or all of set preserving in the request accessed node of user may include that and determines that user needs to read according to the request of user
The version of the metadata taking, the version of metadata being preserved according to multiple nodes and the load of multiple node, by multiple nodes
One of node by need read metadata be supplied to user.
Metadata management method according to embodiments of the present invention, in the case of carrying out write operation to metadata, from main section
Point starts to write, and writes successfully 1,2 or n and returns, the affair mechanism between multinode, if being related to node too much,
Write latency can be greatly increased, affect availability, or can be by user setup minimum successful node number (1 <=minsuc <
=n).
For example, as shown in figure 3, minsuc is set to 1, then when a metadata node (i.e. metadata shown in figure
Storage host node) can return to when writing successfully.The application module of user setup can be data definition, data loading or number
It is investigated that asking.Application module from host node reading writing information, or can read required information from from node.
For example, as shown in figure 4, minsuc is set to 2, then when two metadata node (i.e. shown in Fig. 4 unit
Data storage host node and metadata store from node) just return when all write successfully.In other words, when a metadata
Node writes successfully, and another metadata node writes unsuccessfully, then be carried out rolling back action to two metadata node.User sets
The application module put can be data definition, data loading or data query.Application module can be from two nodes above
Reading writing information, or can read required from the 3rd node and later node (i.e. from node after minsuc node)
Information.Can obtain in the same manner, the operating procedure when minsuc is set to larger than 2 value, for avoiding repeating no longer to describe one by one.
In the case that node is updated, can be according to the predetermined sequence of multiple nodes, to the node needing renewal
Carry out unidirectional renewal successively.
, by the way of unidirectional renewal, renewal can only be from host node for metadata management method according to embodiments of the present invention
Minsuc node before starting sequentially to update, for the node later positioned at minsuc, will be updated by the way of asynchronous.
Unidirectional update mode can guarantee that the latest edition of all of metadata information exists on minsuc fixing node,
Ensure that the metadata consistency of minsuc node of user setup maximum of probability.If user will read up-to-date first number
According to any one that then can be directly connected in front minsuc node is read out, and such system is according to user's request
The loading condition of metadata version and metadata node selects suitable node to initiate metadata request.If the unit that user reads
Data never updates, for example, obtain the letter that partitions of database strategy etc typically will not be changed after importing the data easily
Breath, now may be selected by the node (being not necessarily host node) of most lightly loaded, obtains metadata information.If user's request
Metadata frequent updating then must select the acquisition metadata information of a light load in the node of front minsuc.
From practical application it is contemplated that read operation is frequent, therefore emphasis point is not ensureing all to save for present invention design
The concordance of point, and it is to ensure that host node data is up-to-date, from node is not necessarily up-to-date, and designs unidirectional update mechanism, so
The situation of concurrent update order entanglement can be avoided it is ensured that the data only one of which of latest edition, different nodes update not simultaneously
Can conflict.Realize read and write abruption, it is synchronized update that data is updated for front minsuc node, starts to update from host node again
To other replica node, front minsuc node updates are successfully returned to, and node afterwards adopts asynchronous refresh,.So may be used
Data with minsuc node before ensureing is up-to-date forever, and front minsuc node externally provides up-to-date metadata simultaneously
Information, updates locking granularity and is also only limited on the table in front minsuc node.Read operation is then that all metadata node are same
When provide.So ensure that the high availability of read operation avoids updating the conflict of operation simultaneously, realize final consistency, and will not
Impact metadata system externally provides correct service.
Additionally, according to embodiments of the present invention, there is also the copy memory mechanism of a lot of system design metadata.With hbase
As a example, hbase is data base (the not only structured query of a large-scale distributed non-relational
Language, abbreviation nosql) data base, it adopts the framework of similar b tree, by metadata Bedding storage, and by metadata group
It is made into the form of the internal table structure of hbase, stored metadata on multiple nodes using the copy mechanism as data.By
This can be seen that metadata only describes the information such as storage mode and the storage location of data, therefore, the metadata of mass data
The hundreds of or even thousand of order of magnitude would generally be differed from data itself, common unit just can be competent at the management to metadata and dimension
Shield, the metadata storage mode of hbase only applies to store simple data positional information, and more detailed data feature is then no
Method stores.And can support and realize quick search and renewal using structural data storage system.Hive is (based on hadoop's
One Tool for Data Warehouse) metadata be just stored in structured database, simply the depositing on hdfs of its storage at present
The File Mapping of the specific format of storage is the metamessage of form, and is the mysql data base of unit, does not account for highly reliable
Property.Can also be by data partition strategy, node state and data base in the distributed storage of structural data more than comprehensive
State extracts as independent meta data block storage.Various types of Distributed Storage can be with reference to hive's
Pattern, using relational data library management metadata.This thinking is carried out arranging extension by the present invention, and obtains good in practice
Application effect.
According to one embodiment of present invention, there is provided a kind of managing device of metadata.
As shown in figure 5, the managing device of the metadata being provided according to embodiments of the invention includes:
Configuration module 51, for being multiple set by metadata configurations;
Preserving module 52, for the multiple collection in the Preservation Metadata in the way of full duplication of each node in multiple nodes
Close;
Access modules 53, for the part or all of set preserving in the request accessed node according to user.
Wherein, access modules 53 are further used in the case that node is updated, predetermined according to multiple nodes
Sequence, carries out unidirectional renewal successively to the node needing to update.
This device is equally able to carry out each operation described in said method embodiment, and detailed process is above
It is described, be not repeated herein.
In sum, by means of the technique scheme of the present invention, by the form being multiple set by metadata configurations
It is saved in node, then provides service to user, the resource of metadata can be integrated, optimize the storage organization of metadata, thus
Improve the efficiency that metadata node is updated, and due to employing the mode of full duplication, reduce the general of loss of data
Rate.Additionally, using the complicated metadata information of the higher relational data library management huge number of reliability, relation, using pass
It is the concordance that the instruments such as the affairs of data base, storing process, trigger ensure to update inside metadata, and on this basis
Using following technology: first, multinode repeats to store, and improves metadata memory reliability, even if there being node failure, other section
Point can continue to provide service;Second, multinode provides service simultaneously, and the access shunting of high concurrent is concurrently gulped down so that accessing
The amount of telling greatly improves;3rd, the addition of unidirectional update mechanism, indicate that front minsuc node must be up-to-date metadata.
Above technology further increases high reliability and the high efficiency of metadata management, and then improves distributed data-storage system
Resource utilization.
The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all essences in the present invention
Within god and principle, any modification, equivalent substitution and improvement made etc., should be included within the scope of the present invention.
Claims (7)
1. a kind of management method of metadata is it is characterised in that described management method includes:
Metadata configurations are multiple set;
Multiple set of described metadata are preserved in each node in multiple nodes in the way of full duplication;
According to the part or all of set preserving in the request accessed node of user;
Wherein, host node data is up-to-date, and from node is not necessarily up-to-date, and designs unidirectional update mechanism;
In the case that the part or all of set preserving in a node is updated, if be updated successfully to a set, and
There is, with this set, the set associating and update unsuccessfully, then execute rollback to this set and with improving and there is the set associating;
During node is updated, if updating the number of nodes being successfully updated when stopping to be less than predetermined threshold, to more
The node execution rollback newly completing;
In the case of determining that according to the request of user user needs to read metadata, protect according in the request accessed node of user
The part or all of set deposited includes:
Determine the version of the metadata that user needs to read, the metadata preserving according to the plurality of node according to the request of user
Version and the plurality of node load, by one of the plurality of node node will need read metadata provide
To described user.
2. management method according to claim 1 is it is characterised in that further include:
Configure each set of described metadata by the form of form, each form includes the title of metadata in this set
With corresponding content.
3. management method according to claim 2 is it is characterised in that described content includes at least one of:
The type of metadata, the descriptive information of metadata.
4. management method according to claim 1 it is characterised in that in node preserve partly or entirely gather into
In the case that row updates, it is updated using affair mechanism, in units of gathering.
5. management method according to claim 1 is it is characterised in that further include:
In the case that node is updated, according to the predetermined sequence of the plurality of node, the node that needs are updated is successively
Carry out unidirectional renewal.
6. a kind of managing device of metadata is it is characterised in that described managing device includes:
Configuration module, for being multiple set by metadata configurations;
Preserving module, for preserving multiple collection of described metadata in the way of full duplication in each node in multiple nodes
Close;
Access modules, for the part or all of set preserving in the request accessed node according to user;
Update module, up-to-date for host node data, from node is not necessarily up-to-date, and designs unidirectional update mechanism;
First performing module, in the case that the part or all of set for preserving in a node is updated, if right
One set is updated successfully, and there is, with this set, the set associating and update unsuccessfully, then close to this set and with improving and existing
The set execution rollback of connection;
Second performing module, for during being updated to node, if it is few to update the number of nodes being successfully updated when stopping
In predetermined threshold, then node execution rollback renewal being completed;
Update module is further used in the case of determining that according to the request of user user needs to read metadata, according to user
Request accessed node in preserve part or all of set include:
Determine the version of the metadata that user needs to read, the metadata preserving according to the plurality of node according to the request of user
Version and the plurality of node load, by one of the plurality of node node will need read metadata provide
To described user.
7. managing device according to claim 6 is it is characterised in that described access modules are further used for node is being entered
In the case that row updates, according to the predetermined sequence of the plurality of node, unidirectional renewal is carried out successively to the node needing to update.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310068332.9A CN103106286B (en) | 2013-03-04 | 2013-03-04 | Method and device for managing metadata |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310068332.9A CN103106286B (en) | 2013-03-04 | 2013-03-04 | Method and device for managing metadata |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103106286A CN103106286A (en) | 2013-05-15 |
CN103106286B true CN103106286B (en) | 2017-02-01 |
Family
ID=48314141
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310068332.9A Active CN103106286B (en) | 2013-03-04 | 2013-03-04 | Method and device for managing metadata |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103106286B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103530387A (en) * | 2013-10-22 | 2014-01-22 | 浪潮电子信息产业股份有限公司 | Improved method aimed at small files of HDFS |
CN104199963A (en) * | 2014-09-19 | 2014-12-10 | 浪潮(北京)电子信息产业有限公司 | Method and device for HBase data backup and recovery |
CN104506625B (en) * | 2014-12-22 | 2018-04-17 | 国云科技股份有限公司 | A kind of method for lifting cloud database metadata node reliability |
CN105426439B (en) * | 2015-11-05 | 2022-07-05 | 腾讯科技(深圳)有限公司 | Metadata processing method and device |
CN105824867A (en) * | 2015-12-14 | 2016-08-03 | 广东亿迅科技有限公司 | Mass file management system based on multi-stage distributed metadata |
CN105554135B (en) * | 2015-12-23 | 2019-03-05 | 北京奇虎科技有限公司 | The method and device of synchronous metamessage between back end |
CN106528876B (en) * | 2016-12-09 | 2019-08-23 | 北京奇虎科技有限公司 | The information processing method and distributed information processing system of distributed system |
CN109213760B (en) * | 2018-08-02 | 2021-10-22 | 南瑞集团有限公司 | High-load service storage and retrieval method for non-relational data storage |
CN110134340B (en) * | 2019-05-23 | 2020-03-06 | 苏州浪潮智能科技有限公司 | Method, device, equipment and storage medium for updating metadata |
CN111090803A (en) * | 2019-11-22 | 2020-05-01 | 贝壳技术有限公司 | Data processing method and device, electronic equipment and storage medium |
CN111143027A (en) * | 2019-12-06 | 2020-05-12 | 北京浪潮数据技术有限公司 | Cloud platform management method, system, equipment and computer readable storage medium |
CN111752954B (en) * | 2020-06-29 | 2022-08-12 | 深圳前海微众银行股份有限公司 | Large-scale feature data storage method and device |
CN112231399A (en) * | 2020-09-25 | 2021-01-15 | 中国建设银行股份有限公司 | Method and device applied to graph database |
CN113297173B (en) * | 2021-05-24 | 2023-10-31 | 阿里巴巴新加坡控股有限公司 | Distributed database cluster management method and device and electronic equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101576915B (en) * | 2009-06-18 | 2011-06-08 | 北京大学 | Distributed B+ tree index system and building method |
CN102158546A (en) * | 2011-02-28 | 2011-08-17 | 中国科学院计算技术研究所 | Cluster file system and file service method thereof |
CN102314506A (en) * | 2011-09-07 | 2012-01-11 | 北京人大金仓信息技术股份有限公司 | Distributed buffer management method based on dynamic index |
CN102411637A (en) * | 2011-12-30 | 2012-04-11 | 创新科软件技术(深圳)有限公司 | Metadata management method of distributed file system |
CN102521297A (en) * | 2011-11-30 | 2012-06-27 | 北京人大金仓信息技术股份有限公司 | Method for achieving system dynamic expansion in shared-nothing database cluster |
CN102855284A (en) * | 2012-08-03 | 2013-01-02 | 北京联创信安科技有限公司 | Method and system for managing data of cluster storage system |
-
2013
- 2013-03-04 CN CN201310068332.9A patent/CN103106286B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101576915B (en) * | 2009-06-18 | 2011-06-08 | 北京大学 | Distributed B+ tree index system and building method |
CN102158546A (en) * | 2011-02-28 | 2011-08-17 | 中国科学院计算技术研究所 | Cluster file system and file service method thereof |
CN102314506A (en) * | 2011-09-07 | 2012-01-11 | 北京人大金仓信息技术股份有限公司 | Distributed buffer management method based on dynamic index |
CN102521297A (en) * | 2011-11-30 | 2012-06-27 | 北京人大金仓信息技术股份有限公司 | Method for achieving system dynamic expansion in shared-nothing database cluster |
CN102411637A (en) * | 2011-12-30 | 2012-04-11 | 创新科软件技术(深圳)有限公司 | Metadata management method of distributed file system |
CN102855284A (en) * | 2012-08-03 | 2013-01-02 | 北京联创信安科技有限公司 | Method and system for managing data of cluster storage system |
Non-Patent Citations (1)
Title |
---|
《基于对象存储的新型元数据管理策略》;方圆等;《计算机工程》;20120229;第38卷(第3期);正文第3、4节 * |
Also Published As
Publication number | Publication date |
---|---|
CN103106286A (en) | 2013-05-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103106286B (en) | Method and device for managing metadata | |
US11288252B2 (en) | Transactional key-value store | |
US10754854B2 (en) | Consistent query of local indexes | |
US9659050B2 (en) | Delta store giving row-level versioning semantics to a non-row-level versioning underlying store | |
US9576038B1 (en) | Consistent query of local indexes | |
US11023453B2 (en) | Hash index | |
US8768977B2 (en) | Data management using writeable snapshots in multi-versioned distributed B-trees | |
CN105190623B (en) | Log record management | |
CN105190533B (en) | Snapshot in situ | |
US11841844B2 (en) | Index update pipeline | |
CN107423422B (en) | Spatial data distributed storage and search method and system based on grid | |
US20180011892A1 (en) | Foster twin data structure | |
US11442961B2 (en) | Active transaction list synchronization method and apparatus | |
US20170351543A1 (en) | Heap data structure | |
CN105210062A (en) | System-wide checkpoint avoidance for distributed database systems | |
CN105190622A (en) | Fast crash recovery for distributed database systems | |
CN105408895A (en) | Latch-free, log-structured storage for multiple access methods | |
US11100083B2 (en) | Read only bufferpool | |
CN106575297A (en) | High throughput data modifications using blind update operations | |
US9047019B2 (en) | Shared temporary storage management in a shared disk database cluster | |
US9384257B2 (en) | Providing multiple concurrent transactions on a single database schema using a single concurrent transaction database infrastructure | |
CN115114370B (en) | Master-slave database synchronization method and device, electronic equipment and storage medium | |
CN105956041A (en) | Data model processing method based on Spring Data for MongoDB cluster | |
CN109446258A (en) | A kind of distributed data storage method and system | |
CN115964444A (en) | Cloud native distributed multi-tenant database implementation method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220812 Address after: 100089 building 36, courtyard 8, Dongbeiwang West Road, Haidian District, Beijing Patentee after: Dawning Information Industry (Beijing) Co.,Ltd. Patentee after: DAWNING INFORMATION INDUSTRY Co.,Ltd. Address before: 100193 No. 36 Building, No. 8 Hospital, Wangxi Road, Haidian District, Beijing Patentee before: Dawning Information Industry (Beijing) Co.,Ltd. |