CN107977446A - A kind of memory grid data load method based on data partition - Google Patents
A kind of memory grid data load method based on data partition Download PDFInfo
- Publication number
- CN107977446A CN107977446A CN201711309940.9A CN201711309940A CN107977446A CN 107977446 A CN107977446 A CN 107977446A CN 201711309940 A CN201711309940 A CN 201711309940A CN 107977446 A CN107977446 A CN 107977446A
- Authority
- CN
- China
- Prior art keywords
- data
- grid
- key
- loading
- memory
- 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.)
- Pending
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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24552—Database cache management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2471—Distributed queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Invention is related to a kind of memory grid data load method based on data partition.The distributing position of data is calculated according to grid data partition information, loading tasks are distributed for each grid node so that each node only load store is in local data, so that cluster communication cost and data migration cost are avoided, the performance of lifting data loading.In addition, only loading the key business data by user configuration in loading procedure, memory source is made full use of.
Description
Technical field
The present invention relates to a kind of memory grid data load method based on data partition, belong to software technology field.
Background technology
With the development of network technology, there is volatile growth in data volume and network traffics, and network application is faced with
The demand that Mass storage and high concurrent access.Under the storage of large-scale data and high concurrent read-write business, traditional relational
There are many limitations, non-relational number in mass data storage, high concurrent access and system extension etc. for database
High speed development is able to according to storehouse.Memory data grid is as a kind of distributed memory data object management middleware, internal storage data net
Lattice mainly have following features：（1）Flexible storage model：Stored based on key-value, user can flexibly define value
Inside the Nomenclature Composition and Structure of Complexes without data storage should be influenced, and close association is had no between the storage of data, had very high
Autgmentability, have in terms of mass data is handled and have great advantage.Meanwhile memory data grid supports distributed queue, distribution
The data structures such as formula set, further improve the flexibility of its data storage, can largely adapt to existing data
Storage demand；（2）The cluster of dynamic expansion：The onrelevant stored due to the equity and data of each node of memory data grid
Property, the node for the additions and deletions memory data grid that user can be elastic.Meanwhile the efficient data back mechanism of each node is its dynamic
Extension provides high reliability；（3）Efficient memory storage：Memory data grid is stored data in memory, avoids magnetic
The low bottleneck of disk I O access performance, the performance that the storage of significant increase data accesses, can preferably tackle under big data background
High concurrent data access request.
Memory data grid still suffers from some shortcomings in data loading, the fusion of interior external memory, data access interface etc., main
Show the following aspects：（1）Data loading performance is low：Memory data grid is based on memory storage, is opened for the first time in system
, it is necessary to which data are loaded into memory data grid from relevant database when dynamic.When tackling large-scale data scene, how
Efficiently critical data is loaded into the memory data grid of multinode composition, many memory data grid schemes should in reply
Still lack complete scheme during problem, loading performance is integrally relatively low；（2）Interior external memory fusion is not perfect enough：Memory data grid with
Relevant database is relatively independent, and the consistency problem of both data is the key of system application.The internal storage data of mainstream at present
Mesh products are read data in memory data grid from disk by Read-Through modes, while pass through Write-
Data are persisted in disk by Through/Write-Behind modes from memory data grid by either synchronously or asynchronously mode
（JI S, WANG W, YE C, et al. Constructing a data accessing layer for in-memory
data grid; proceedings of the Proceedings of the Fourth Asia-Pacific
Symposium on Internetware, F, 2012 [C]. ACM.）, or by application program oneself come safeguard caching and
The data consistency of database（GAUR N, KAPLINGER T E, BHOGAL K S, et al. Dynamic map
template discovery and map creation [M]. Google Patents. 2013.）.But when cannot feel
When knowing that the third-party application of memory data grid directly updates backstage disk database, due to memory data grid and using journey
Sequence can not perception data change, easily there are out-of-date cache problem（GWERTZMAN J, SELTZER M I. World
Wide Web Cache Consistency; proceedings of the USENIX annual technical
conference, F, 1996 [C].）, interior external memory fusion needs further perfect.（3）Lack unified data access to connect
Mouthful：Memory data grid has very high flexibility in terms of data storage, but flexible data storage result in data access and connect
The disunity of mouth.The memory data grid product of mainstream lacks the compatibility to Legacy System mostly at present, particularly to structuring
The compatibility of query language.Although mainstream memory data grid product support part SQL syntax, due to the SQL languages of its support
Method still imperfection, there is some difference with traditional database development pattern for data access mode, this is largely limited
The further development of memory data grid, developer need to be directed to specific business and specific memory data grid product
Secondary development is done, adds extra exploitation cost and learning cost.
The content of the invention
The purpose of the present invention：By extending the support to JDBC interfaces, SQL language and the improvement of SQL request process flow,
Further lift the Data Access Integration ability of memory data grid.
The principle of the present invention：Loaded in parallel scheme based on data partition information.It is automatic according to grid data partition information
Loading tasks are distributed for each node, only load local data, avoid cluster communication and data migration cost, lift data response rate
Can, in addition, by the filtering to non-critical data, lift memory usage.
In memory data grid initial start-up, it is necessary to by data from being loaded into memory data grid from the background, so that using
The data of memory data grid, the performance of lifting system can directly be accessed.Data loading is one in system initialization process
Item important process, efficiently realizes that a data loading scheme can effectively lift the performance of grid initialization.
The problem of one complete data loading scheme needs to consider following aspects：（1）Which data loaded：From the background
The different data of each business are stored in database, should only loading be moved in data load process in memory data grid
How service related data, effectively define critical data, filters out incoherent data, is that resource effect is lifted in loading procedure
One of key of rate and system performance；（2）Data model translation：Memory data grid stores data in the form of key/value, has
Beneficial to the extending transversely of grid node, wherein key and value are data object, and in relevant database, data are with table
Form stored, therefore to say that data are loaded into memory data grid from relevant database, first have to implementation relation type
Automatic conversion of the data model to key/value data models；（3）How the loading tasks of grid node are distributed：Internal storage data
Grid is a distributed system that can be extending transversely, and the data of each node storage do not have obvious relevance, are loaded in data
During, how for each node distribute loading tasks, efficiently utilize the cpu resource and memory source of each node so that whole
The data loading efficiency highest of a grid, is another key of data loading.
The first step, data model translation.
Invention realizes the data model automatic switching method in data load process, mainly includes the generation of key,
The generation of Value and the foundation of index, in the generation of key, mainly consider two problems：（1）The uniqueness of key：Due to interior
Deposit data grid data generally exists with the structure of Map, and the uniqueness of key is the necessary condition of Map storage data；（2）Key's
Ease for use：In memory data grid, the cryptographic Hash of key can be calculated so that it is determined that the specific distribution of data, a rational key are needed
Ensure the easy calculating of cryptographic Hash, meanwhile, the digital independent of memory data grid is all based on key, if the generation of key is excessively
Complexity, unnecessary performance cost can be all brought in digital independent.In order to ensure the uniqueness of key and ease for use, in invention
Key exist all in the form of character string, for the single major key situation in database, directly choose the major key in database table
As the key of data object, for more major key situations in database table, multiple row major key is accorded with into combination by special interval and is spelled
Connect, generate single key, be each using major key autoincrement mode to the situation for not having clearly to specify major key in database table
One integer of value object maintenances is from variable is increased as corresponding key values.
In the generation of value, what emphasis need to consider is the versatility of value.Due to different tables in relevant database
Data contain different structures, how different table data are uniformly generated to the value in Map, be emphasis consider the problem of.Hair
It is bright by a database table Mapping and Converting into a Map, one group of key/ in the corresponding Map of a record in database table
Value key-value pairs.Metamessage by database table is each one data object class of table dynamic generation, the type of attribute in class
Corresponded with the type of relevant database attribute, all data object classes contain same parent, in this way, being database table
Each object instance of every record generation, in the Map that same structure can be stored.
Index information in relevant database is to provide the correct means of efficient data access.Invention is in order to by relationship type
The index information of database is synchronized to memory data grid, specially devises index manager.Internal storage data is synchronized in data
Before grid, the metamessage for beginning with database table creates corresponding distribution Map, while can extract database column index
Information, adds corresponding index in Map.In the query process based on index, adding the Map of index can be arrived with immediate addressing
Corresponding value objects.
Second step, data homogenization loading.
When original single node loading scheme is applied to memory data grid cluster, due to there was only host node in loading number
According to, and data are re-distributed to respectively from node, single node bottleneck is so readily formed, and all tasks are all in host node
Perform, cause the non-uniform situation of the utilization of resources, therefore, invention design realizes data homogenization loading scheme.
In data homogenize loading scheme, memory data grid elects master nodes first, and master nodes are read
Relevant database metamessage is taken, and is injected into memory data grid, each slave nodes read metamessage and turn according to model
The method of changing automatically generates local class, then master nodes reading database data amount information, and by the uniform burst of data, by number
Corresponding according to burst information and node, master nodes distribute each node loading tasks, and each node loads respective data fragmentation
Data, complete data homogenization loading scheme.
3rd step, service template design
Loaded in parallel scheme based on data partition information can be by the configuration of key business template, to load and key business phase
The data of pass, so that data are loaded with the flexibility of more height, while the performance of lifting system loading and the profit of memory
With efficiency, invention mainly configures key business data from following three granularity.
（1）According to database table name：By configuration file configuration and the relevant database table name of key business, in data plus
During load, only reading can directly be filtered with the relevant database table of key business, incoherent table, will not be loaded into memory
In data grids.
（2）According to field type：In relevant database, the field containing many specific types, for example Blob, Clob
Deng typical Blob is a pictures or an audio files, due to their size, must be used in relevant database
Special mode is handled, and Clob fields are generally big character object, such as XML document.Due to these fields spy of itself
Different attribute, is loaded into meeting extreme influence performance after memory.In the design of service template, user can set the mistake of type field
Filter, filtering out these influences the data of memory storage efficiency and access performance, lifting system performance.
（3）Filtered according to SQL：User can customize data query sentence, and in data load process, loading scheme can be certainly
It is dynamic to load specified data, rather than global data according to user-defined query statement.
4th step, the loaded in parallel based on data partition information.
To solve the deficiency that memory data grid available data homogenizes loading scheme, the property of data loading is further lifted
The loaded in parallel scheme based on data partition information can be proposed with the service efficiency of memory, invention, its feature is mainly manifested in：
（1）Master nodes efficiently distribute loading tasks by grid data partition information for each grid node so that each node only adds
Carry and be stored in local data after passing through uniformity Hash calculation, so as to avoid data from migrating cost and cluster communication in each node
Cost, lifts data loading performance；（2）Only loading with the relevant data of key business, key business can from advance it is designed
Service template configures.
In order to reduce cost on network communication and data migration cost in data homogenization loading scheme, invention proposition is based on
The loaded in parallel scheme of data partition information.In the loaded in parallel scheme based on data partition information, start memory number first
Go out master nodes according to grid cluster, and by cluster election, master nodes obtain distributed lock；Master nodes read relation
The major key information of type database data, according to grid data partition information, major key information is distributed into different set makes total
According to being stored in local node after loading, and release profile formula is locked；Each grid node obtains database metadata, and dynamic generation
Local class；All grid nodes concurrently obtain own node from distributed data structure to be needed to load the major key information of data,
Load logic is generated according to major key information, concurrently loads data.
The present invention has the following advantages that compared with prior art：The distribution position of data is calculated according to grid data partition information
Put, loading tasks are distributed for each grid node so that each node only load store is in local data, so as to avoid cluster communication
Cost and data migration cost, the performance of lifting data loading.In addition, the crucial industry by user configuration is only loaded in loading procedure
Business data, make full use of memory source.
Brief description of the drawings
Fig. 1 is memory data grid system architecture.
Embodiment
Below in conjunction with specific embodiments and the drawings, the present invention is described in detail, as making for present invention method
With environment, as shown in Figure 1.
By to the loaded in parallel scheme based on data partition information, delta data catch mechanism and Data Access Integration skill
The research of art, invention realize the memory data grid system towards Cache-Aside patterns on the basis of memory data grid,
Its system architecture diagram and application scenarios.System architecture illustrates the module composition and associated component of internal system, and each module
Request under specific business scenario accesses flow relation and data flow relation, and the operation mechanism of system is described more detail below.
First, each node of memory data grid is started, after a node meets user-defined cluster condition, cluster is opened
Begin loading data, and under cluster environment, the data loading scheme of acquiescence is the loaded in parallel scheme based on data partition information, should
Loading tasks can be distributed for each node automatically under scheme, the data of loading can be distributed to local section according to uniformity hash algorithm
Point, without extra cluster communication and data migration cost.Meanwhile meeting automatic fitration is fallen and user-defined industry in loading procedure
It is engaged in unrelated data, makes full use of memory source, the performance of the loading of lifting system initialization to the full extent.
Then, start CDC components, capture database data change in real time, and change data are transmitted to internal storage data net
Lattice, keep memory data grid data and the data consistency of backstage relevant database.
Finally, JDBC drivings are replaced database-driven in original operation system by user, start application server, user
During the SQL separators that request is driven by the JDBC of internal storage data webmaster, if the request memory data grid is supported and belongs to use
The customized business in family, then transfer to memory data grid to perform, otherwise directly hand to by SQL compilers and enforcement engine etc.
Database performs.The SQL request for transferring to memory data grid to perform, if so that grid data changes, by synchronization or
Person's asynchronous system is persisted to background data base；And the SQL request for transferring to database to perform, if so that user-defined data are sent out
Change is given birth to, then CDC components can capture the data and change and be synchronized to memory data grid, interior under Cache-Aside patterns
Deposit data grid and database are a kind of mechanism of bi-directional synchronization, the further perfect integration program of interior external memory.
The system architecture is a kind of framework of interior external memory fusion, make use of to a certain extent memory data grid it is high simultaneously
Memory data grid, can preferably be fitted in existing system by volatility, be the once new trial of memory data grid system.
The specific implementation of each key technology is discussed below.One complete data loading flow includes metadata loading, model
Conversion, index loading and data loading.The detailed process of data loading is described in detail below.
Metadata loads, i.e. the realization of loadMetaData () method.Database is connected first, and passes through JDBC's
The getMetaData methods of ResultSet classes obtain the metamessage of database data table, and metamessage then is converted to work of increasing income
Have the BeanGenerator objects in cglib.Cglib is powerful a, high-performance, and the Code of high quality generates class libraries, it
Java class can be extended in the runtime and realizes Java interfaces, many well-known Open-Source Tools such as Spring, Hibernate etc.
The dynamic generation of bytecode is realized with it.While metamessage is converted to BeanGenerator objects, metamessage is deposited
Store up in TableMeta classes, to store the extraneous informations such as major key, BeanGenertor is mainly used for class in following model conversion
Dynamic generation, and TableMeta is mainly used for providing the major key information needed in data load process, table name, row name etc. additionally
Information.
The realization of model conversion, that is, populateBeanMap () method.Model conversion process is mainly loaded using metadata
The BeanMap classes in metamessage and Cglib in the BeanGenerator objects and TableMetaData objects that obtain afterwards, come
Dynamic construction class memory data grid Object, the parent of such all classes of data entities.Index loading is loading of databases
The index information of table.Since memory data grid realizes index manager, in data load process, can be connect by JDBC
The index information of mouth extraction database, corresponding index is added in Map, improves the performance of Value inquiries.
After the completion of metadata loading, model conversion and index loading all, the loading of database data is finally carried out, in base
In the loaded in parallel scheme of data partition information, metadata loading, model conversion and index are loaded as described above, specific
In data load process, each node is directed to the data of database, and judging the data, whether the storage is to local, if it is, plus
Carry, so that the data for ensureing to load, which are all storages, arrives local data.Due to being stored in the data routing table of memory data grid
The partition information of cluster information and each node, can be easy to subregion where judging it, so as to determine whether it according to data Id
Place node.
Claims (1)
1. method characteristic is to realize that step is as follows：
The first step, data model translation：The main generation for including key, the generation of Value and the foundation of index, key is with character
The form of string exists, and for the single major key situation in database, directly chooses the major key in database table as data object
Key, for more major key situations in database table, multiple row major key is accorded with by special interval combined and spliced, generated single
Key, is each value object maintenances one using major key autoincrement mode to the situation for not having clearly to specify major key in database table
A integer is from variable is increased as corresponding key values；
Second step, data homogenization loading：Memory data grid elects master nodes first, and master nodes read relation
Type database metamessage, and memory data grid is injected into, each slave nodes read metamessage and according to model conversion methods
Local class is automatically generated, then master nodes reading database data amount information, and by the uniform burst of data, by data fragmentation
Information and node are corresponding, and master nodes distribute each node loading tasks, and each node loads the data of respective data fragmentation, complete
Loading scheme is homogenized into data；
3rd step, service template design：Loaded in parallel scheme based on data partition information can matching somebody with somebody by key business template
Put, come load with the relevant data of key business so that data are loaded with the flexibility of more height, while lifting system
The performance of loading and the utilization ratio of memory, invention mainly come from three granularities such as database table name, field type, SQL filterings
Configure key business data；
4th step, the loaded in parallel based on data partition information：Start memory data grid cluster first, and gone out by cluster election
Master nodes, master nodes obtain distributed lock；Master nodes read the major key information of relevant database data, root
According to grid data partition information, major key information is distributed into different set so that being stored in local node after data loading, and
Release profile formula is locked；Each grid node obtains database metadata, and dynamic generation local class；All grid nodes concurrently from
Own node is obtained in distributed data structure to be needed to load the major key information of data, and load logic is generated according to major key information,
Concurrently load data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711309940.9A CN107977446A (en) | 2017-12-11 | 2017-12-11 | A kind of memory grid data load method based on data partition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711309940.9A CN107977446A (en) | 2017-12-11 | 2017-12-11 | A kind of memory grid data load method based on data partition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107977446A true CN107977446A (en) | 2018-05-01 |
Family
ID=62009940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711309940.9A Pending CN107977446A (en) | 2017-12-11 | 2017-12-11 | A kind of memory grid data load method based on data partition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107977446A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108733781A (en) * | 2018-05-08 | 2018-11-02 | 安徽工业大学 | The cluster temporal data indexing means calculated based on memory |
| CN111523002A (en) * | 2020-04-23 | 2020-08-11 | 中国农业银行股份有限公司 | Main key distribution method, device, server and storage medium |
| CN111984696A (en) * | 2020-07-23 | 2020-11-24 | 深圳市赢时胜信息技术股份有限公司 | Novel database and method |
| CN112667859A (en) * | 2020-12-30 | 2021-04-16 | 北京久其软件股份有限公司 | Data processing method and device based on memory |
| CN112698957A (en) * | 2021-02-02 | 2021-04-23 | 北京东方通科技股份有限公司 | Data processing method and system based on memory data grid |
| CN114398105A (en) * | 2022-01-20 | 2022-04-26 | 北京奥星贝斯科技有限公司 | Method and device for loading data from distributed database by computing engine |
| CN115096291A (en) * | 2022-07-06 | 2022-09-23 | 上海电气集团智能交通科技有限公司 | Electronic map dynamic splicing method based on magnetic nail track |
| CN117435594A (en) * | 2023-12-18 | 2024-01-23 | 天津南大通用数据技术股份有限公司 | Optimization method for distributed database distribution key |
-
2017
- 2017-12-11 CN CN201711309940.9A patent/CN107977446A/en active Pending
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108733781A (en) * | 2018-05-08 | 2018-11-02 | 安徽工业大学 | The cluster temporal data indexing means calculated based on memory |
| CN108733781B (en) * | 2018-05-08 | 2021-10-29 | 安徽工业大学 | Cluster temporal data indexing method based on memory calculation |
| CN111523002B (en) * | 2020-04-23 | 2023-06-09 | 中国农业银行股份有限公司 | Main key distribution method, device, server and storage medium |
| CN111523002A (en) * | 2020-04-23 | 2020-08-11 | 中国农业银行股份有限公司 | Main key distribution method, device, server and storage medium |
| CN111984696A (en) * | 2020-07-23 | 2020-11-24 | 深圳市赢时胜信息技术股份有限公司 | Novel database and method |
| CN111984696B (en) * | 2020-07-23 | 2023-11-10 | 深圳市赢时胜信息技术股份有限公司 | Novel database and method |
| CN112667859A (en) * | 2020-12-30 | 2021-04-16 | 北京久其软件股份有限公司 | Data processing method and device based on memory |
| CN112698957A (en) * | 2021-02-02 | 2021-04-23 | 北京东方通科技股份有限公司 | Data processing method and system based on memory data grid |
| CN112698957B (en) * | 2021-02-02 | 2024-02-20 | 北京东方通科技股份有限公司 | Data processing method and system based on memory data grid |
| CN114398105A (en) * | 2022-01-20 | 2022-04-26 | 北京奥星贝斯科技有限公司 | Method and device for loading data from distributed database by computing engine |
| CN115096291A (en) * | 2022-07-06 | 2022-09-23 | 上海电气集团智能交通科技有限公司 | Electronic map dynamic splicing method based on magnetic nail track |
| CN117435594A (en) * | 2023-12-18 | 2024-01-23 | 天津南大通用数据技术股份有限公司 | Optimization method for distributed database distribution key |
| CN117435594B (en) * | 2023-12-18 | 2024-04-16 | 天津南大通用数据技术股份有限公司 | Optimization method for distributed database distribution key |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107977446A (en) | A kind of memory grid data load method based on data partition | |
| US7984043B1 (en) | System and method for distributed query processing using configuration-independent query plans | |
| US11893046B2 (en) | Method and apparatus for implementing a set of integrated data systems | |
| KR102177190B1 (en) | Managing data with flexible schema | |
| US11093459B2 (en) | Parallel and efficient technique for building and maintaining a main memory, CSR-based graph index in an RDBMS | |
| US5764977A (en) | Distributed database architecture and distributed database management system for open network evolution | |
| JP2019515377A (en) | Distributed Datastore Versioned Hierarchical Data Structure | |
| US20010016843A1 (en) | Method and apparatus for accessing data | |
| CA3137857A1 (en) | Multi-language fusion query method and multi-model database system | |
| US10417257B2 (en) | Non-blocking database table alteration | |
| US6941309B2 (en) | Object integrated management system | |
| US12135707B2 (en) | Maintaining data separation for data consolidated from multiple data artifact instances | |
| US11886411B2 (en) | Data storage using roaring binary-tree format | |
| US20240004853A1 (en) | Virtual data source manager of data virtualization-based architecture | |
| US11960616B2 (en) | Virtual data sources of data virtualization-based architecture | |
| CN105468793A (en) | Automated management method for simulation model data | |
| WO2023151239A1 (en) | Micro-service creation method and related device | |
| CN113468149B (en) | Data model development platform | |
| US11263026B2 (en) | Software plugins of data virtualization-based architecture | |
| CN117056305A (en) | Construction method, model, database system and medium of multisource isomorphic database | |
| US20220374424A1 (en) | Join queries in data virtualization-based architecture | |
| Cobbs | Persistence Programming: Are we doing this right? | |
| CN117828127B (en) | Tree-level cluster user management method based on semi-structured storage | |
| Gilon et al. | Compact deterministic distributed dictionaries | |
| Kpekpassi et al. | PREPRINT NoSQL databases: A survey |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180501 |
|
| WD01 | Invention patent application deemed withdrawn after publication |