CN102880678A - Embedded real-time memory database - Google Patents
Embedded real-time memory database Download PDFInfo
- Publication number
- CN102880678A CN102880678A CN2012103342726A CN201210334272A CN102880678A CN 102880678 A CN102880678 A CN 102880678A CN 2012103342726 A CN2012103342726 A CN 2012103342726A CN 201210334272 A CN201210334272 A CN 201210334272A CN 102880678 A CN102880678 A CN 102880678A
- Authority
- CN
- China
- Prior art keywords
- database
- data
- memory
- index
- real
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the field of application system data management and particularly relates to an embedded real-time memory database with high data reading-writing speed for storing data into a memory. The embedded real-time memory database comprises a database core, a database control management module and a database interface, wherein the database core is used for defining and organizing the data; the database interface is used for accessing the database; and the database control management module is used for controlling a state of the database. The embedded real-time memory database is different from the conventional database; the database is established in the memory, so that the data access speed is greatly accelerated, the dependence on the disk size is reduced, and the data access performance requirements of occasions with high real-time performance are met. Due to the introduction of a two-layer memory concept and a rollback mechanism, the completeness, the uniformity and the safety of the data are guaranteed, and the hardware investment and system maintenance costs can be directly saved by using the memory database while the working efficiency of the database is improved.
Description
Technical field
The invention belongs to the application system data management field, be specifically related to a kind of reading and writing data speed height, save the data in the embedded real-time internal memory database in the internal memory.
Background technology
Along with the widespread use of embedded system and constantly popularizing of embedded real-time operating system, embedded system has become the core technology of a lot of industry developments.Enter 2012, intelligent terminal, Internet of Things, cloud computing industry have quietly entered the stage of high speed development, a large amount of data analyses is processed and has been transferred under the embedded environment from traditional PC, and the data management problem under the embedded environment becomes the important step in the system.Make up the embedded real-time data base system and become problem needing to overcome in the embedded development.
Because there is very big difference in the embedded real-time data base system with common Enterprise Data base management system on running environment and the method for operation, it is too huge to be mainly manifested in traditional enterprise-level database such as Oracle, DB2, cpu dominant frequency and disk size there are very high requirement, can't on the limited embedded system of hardware resource, install and use; In addition, traditional database lacks the mechanism of real-time response, and is subject to R/W speed of hard disc, can't satisfy the requirement of high real-time, and in the very high control system of some requirement of real-times, it is helpless that traditional database more seems.Traditional disk database need to be accessed the operation that disk carries out data continually, owing to will carry out on the one hand the Mechanical Moving of magnetic head to the operation of disk read-write data, being subject on the other hand system call (interrupts finishing by CPU usually, be subject to the restriction in cpu clock cycle) impact of time, when data volume very large, when frequent operation and complexity, will expose a lot of problems.The memory database of the present invention's design has been abandoned the traditional approach of data in magnetic disk management, all in internal memory, redesigned architecture based on total data, and aspect data buffer storage, fast algorithm, parallel work-flow, also carried out corresponding improvement, greatly improve the access speed of data, satisfy the requirement of high real-time occasion data access performance.
Meanwhile present existing embedded database mainly is to be based upon above the Flash, because the Flash read or write speed is slower, and exists the problem of erasing and writing life, so present most embedded database performance is unsatisfactory; The supported data rollback is not machine-processed for present embedded database in addition, the maloperation corrupt data occurs easily, and lacks good data protection mechanism, data file occurs easily and damages.The memory database of the present invention's design combines Flash with memory database, a large amount of data manipulation (insert, delete, revise, inquire about) is placed in the internal memory, greatly reduce the access times to Flash, increased the serviceable life of Flash when improving access speed.Adopt simultaneously data buffer storage rollback mechanism, guaranteed data's consistency, effectively prevented the maloperation corrupt data.Further in order to prevent loss of vital data, adopted the mode of two layers of storage, be kept in the internal memory for real time data, preserved synchronously to Flash according to data importance priority and timing.
Summary of the invention
The object of the present invention is to provide a kind of built-in field that solved to use traditional large database problem based on disk, improved the access speed of data, reduction is satisfied the embedded real-time internal memory database of the data access performance requirement of high real-time occasion to the dependence of disk size.
The object of the present invention is achieved like this:
A kind of embedded real-time internal memory database, comprise database core, database control and management module, three modules of database interface, database core defines and organizes data, database interface carries out access visit to database, and the Data Control administration module is controlled the state of database.
Database core adopts relational data model, the tables of data of database core adopts the bivariate table structure, every bivariate table is used for creating major key index and the secondary index that can accelerate bivariate table record retrieval speed, and database control and management module is by the load-on module that powers on, regularly the module of depositing forms.
Database control and management module is divided into the loading that powers on, two submodules of depositing regularly.
Memory database adopts the mode of double-deck storage, namely has real-time data in the memory table, and the data in regularly will showing are kept in the physical hard disk, after the database power down, restarts and can reload data from hard disk.
Every table all can create index, comprises that a major key index and quantity are greater than one secondary index.
Beneficial effect of the present invention is:
The present invention is different from traditional database, and building database has improved the access speed of data greatly in internal memory, reduces the dependence to disk size, and satisfies the data access performance requirement of high real-time occasion.Introduce concept and the rollback mechanism of two layers of storage, guaranteed integrality, consistance and the security of data, when promoting database work efficient, used memory database can directly save hardware investment and system maintenance cost.
Description of drawings
Fig. 1 is the database loading schematic flow sheet that powers on;
Fig. 2 is the database access schematic flow sheet;
Fig. 3 is that database is realized example schematic.
Embodiment
Database core of the present invention adopts relational data model, and the tables of data of database core adopts the bivariate table structure, and every bivariate table is used for creating major key index and the secondary index that can accelerate bivariate table record retrieval speed.Each table, index and queue object can instantiations, and each table example can be associated with index, formation, and the method function that the table example has oneself is realized specific function.Table adopts relevant database bivariate table structure commonly used, and every table all can create a plurality of index, one of them major key index, a plurality of secondary indexes.The purpose of setting up index is in order to accelerate the retrieval rate to the tuple of bivariate table.Index object is to describe a kind of data object of the index of dissimilar and algorithm.The concordance list of a specific bivariate table is exactly an instantiation of index object.Index adopts sequential index, hash, a hash hash, b-tree indexed.Database core provides rollback mechanism simultaneously, opens up a block cache log history service data, can carry out data rewind when error in data occurring.
Sequential index: index entry is according to the key word ordering, so can use binary chop when searching, average recall precision is log2N.When this index is increased deletion action, owing to will safeguard original order, will inevitably relate to a large amount of index entries moving in internal memory, so that the type is not suitable for data volume is large, the bivariate table of frequent operation index.
A HASH index: the calculating that this index type needs only by a hash function when retrieval just can navigate to index entry, by the subscript location, once searches, and is fastest.The distribution of its index value has determined the scale of this array, so require index regular, index value distributes compact.The index irregular for index, that index value distributes and disperses, the waste that will produce memory headroom.
Hash HASH index: the combination by modular space and Hash bucket efficiently solve index entry in the sequential index in internal memory the problem of moving and hash index in the problem of memory headroom waste, its essence is that the consumption in the consumption of time and space obtains a half-way house.
B-tree indexed: B tree is proper data structure for disk, because he is a wide and shallow tree, searches a number and need to access seldom node.Most of Database Systems are set with the mutation B+ of a B tree, and he preserves all data on the leaf node of tree.Yet for memory database, the B tree is more suitable than the B+ tree, is not advantage because preserve all data in internal memory at leaf, and he is wasting space too, and storage efficiency is too low.The memory usage of B tree is reasonable, so he is proper for memory database; Search speed is than very fast (when use binary chop, only the seldom a part of node of access), and renewal speed is also than comparatively fast, good operation performance, and Dynamic Maintenance is conveniently.
Its assembly that comprises mainly is:
1) leaf node: comprise the data line in the direct Compass of clauses and subclauses.
2) branch node: other branch node or leaf node in the clauses and subclauses sensing index that comprises.
3) root node: a b-tree indexed only has a root node, and it is actual to be exactly the branch node that is positioned at the top of tree.
Unidirectional formation and dique are used for some resources that the maintenance data base system needs.
Database control and management module is divided into the loading that powers on, two submodules of depositing regularly.Memory database adopts the mode of double-deck storage, the storage organization of system is divided into two levels, ground floor is internal memory, be memory database, as basic-level support, it is the key of real-time dataBase system to the high performance requirements of whole real-time dataBase system with memory database, be used for the processing of program operation and real time data, its access speed is fast, need not to carry out the I/O operation of disk, therefore is suitable for most management and the operation of real time data; The second layer is external memory, usually adopts some permanent storage appliances, need to read and write the I/O operation, is used for the historical data of storage system.Like this, be placed in the external space for the relatively low data of those changeless data (such as the engineering unit of analog quantity) or frequency of access (such as data backup or the Log backup that only just uses when the fault recovery), and the data of all real time datas or work at present part are resided in the internal memory, avoid the operation of database file, greatly promoted the performance of real-time dataBase system.According to the data significance level, for each tables of data arranges priority, when having real-time data in the memory table, the data during timing will be shown according to priority are kept in the physical hard disk, after the database power down, restart and can reload data from hard disk.
Database interface provides the approach in user accesses data storehouse.In the situation of the same database of a plurality of user concurrents ground access, can guarantee consistance and the integrality of database data.Consumer process is realized by sending request to the database access process the operation of database, multi-user's Concurrency Access can be converted like this to the serial access for the database access process.
Definition for field in the database table is specified by data dictionary, and data dictionary has comprised the definition of all data types in the database, comprises the common data types such as shaping, character type, floating type.The capacity of data form can be configured by configuration file, and can join max cap. will determine according to the Installed System Memory size.The static allocation mode is adopted in the databases management, and purpose is to improve Memory Allocation efficient, strengthens real-time.Internal memory is divided into some memory pools, memory pool inside is further subdivided into some memory blocks that vary in size, adopt the mode of size coupling during the application internal memory, look for first the minimum internal memory that can match.Can reduce in this way to reduce the expense of dynamically opening up internal memory, and improve the reliability of system.
When starting, database at first creates tables of data according to data dictionary for the size of the definition of field and table capacity, the character type field that for example to comprise a length in form be 100 bytes, the capacity of this table is 1000, and this table memory size that need to take is 100 to take advantage of 1000 to be the 100K byte so.Then data are loaded into the internal memory data creation index of again loading being finished from Flash.
Need to determine according to the multiplicity of table record key word the size of conflict bucket for the Hash index, for example exist two record data key words to be respectively 2 and 12, be mould if get 10, their remainder all is 2 so, and these two records can be stored in the conflict bucket like this; For the branch node piece, its directory entry that comprises is all arranged in order for b-tree indexed.Each directory entry has two fields.The first character segment table shows the minimum key value that comprises in the index block that links below current this branch node piece; Second field is four bytes, the address of the index block that expression links, and a following index block is pointed in this address.The record line number that can hold in a branch node piece is determined by the length of data block size and index key assignments.For the leaf node piece, its directory entry that comprises is the same with branch node, all is (default is that ascending order is arranged, and also can be appointed as descending sort when creating index) of arranging in order.Each directory entry (also can be called every record) also has two words.For each index block, default PCTFREE is 10%, and that is to say at most can only section.The first character segment table shows the key assignments of index, is a value for single-row index; For the multiple row index then be a plurality of values be used in combination wherein 90%.Simultaneously after the 9i, this also can not use up in 90%, can only use wherein about 87%.That is to say that the space that can actually be used for depositing index data in the data block of 8KB is approximately 6488(8192 * 90% * 88%) individual byte.
The data volume of normal conditions form is larger, because the reading speed of disk or electric board is slower, once loading mass data, to internal memory CPU to be taken too high, therefore powers on to load employing and load in batches and postpone loading technique, according to the priority of data form, at first load the high table of priority.The rollback buffer zone of initial creation 10M size, afterwards according to the big or small dynamic change of Installed System Memory, when out of buffers again from the Installed System Memory application, can be with a part of exchanges data in Flash when low memory.
Regularly the data in the memory database are synchronized in the disk by regularly deposit mechanism, prevent database power down obliterated data, this flow process as shown in Figure 1.
When transaction failure recovers, tuple attributes can leave in two memory address spaces, its current image is placed in the address space of " effectively " memory address indication, be the memory address space of significance bit institute mark. in order to improve the concurrency of affairs operation, real-time dataBase system adopts optimistic control method usually. therefore, another memory address space deposit in have two kinds may: the one, when affairs had been submitted to and do not write permanent storehouse, what deposit was the before-image of property value; The 2nd, when affairs with its modification but affairs when not submitted to, what deposit is without the new attribute of submitting to. when affairs will read a certain attribute in the tuple, check first the significance bit of this attribute, find the address of property value according to the value of significance bit, read property value according to this address again. when affairs will be write a certain attribute in the tuple, find the address of corresponding property value according to the value of significance bit, in this address space, write new value.
Affairs are submitted to and are died young, and seem very simple under this memory organization structure. when the detection of carrying out when affairs finish was not found to conflict, affairs were submitted to, only need significance bit is got non-, and with the location revision 1 of tuple; When affairs were died young, significance bit was constant, and non-effective memory address is put sky, and the respective tuple item in the property value structure is removed.
During system failure recovery, when the routine data base management system conducts a survey, it (is that current affairs execute that database must remain static, and do not have new affairs to begin to carry out). this mode is for being inappropriate to the very high real-time system of time requirement, the time-delay that the very large checkpoint operation of workload causes may make some urgent affairs die young owing to overtime. therefore, in real-time data base, must use activity and the checkpoint of affairs can be operated the algorithm that carry out simultaneously.
Concentrate when the work that we write external memory with the internal memory working version is placed on the checkpoint operation and carry out. in the checkpoint stage, when affairs are attempted Update Table storehouse section, its processing mode can be the same with non-checkpoint section, in the checkpiont algorithm, memory database should be write external memory, the daily record that will not write again external memory writes external memory.
When the system failure occured, whole rejuvenation comprised refitting Reloaded and recovers two steps of Recover.Wherein reset, recovery and issued transaction can carry out simultaneously, restart affairs after needn't waiting memory database to recover fully. system's refitting, order according to affairs priority, its needed database segment of packing into, the most urgent affairs can be moved the earliest. Recovery processing carries out in issued transaction, it also is the pressing degree according to affairs, triggers operation when current affairs T wants data in the access memory database, makes urgent affairs desired data that the right of priority restores be arranged.
Because in memory database, the data modification of not submitting affairs to is can not write external memory. therefore, system only needs according to daily record when recovering, and affairs are REDO and need not be UNDO.
The database access process is responsible for processing the access to database, can support simultaneously 1000 Concurrency Access to database.The database access process provides two kinds of access mechanisms, synchronization of access and asynchronous access.This browsing process as shown in Figure 2.
When adopting method of synchronization visit data, data access can be suspended until database returns the needed data of caller.This mode is fit to the caller needs and obtains critical data, is waiting for for hanging up before obtaining data.
When adopting the asynchronous system visit data, data access need not to be suspended, and caller can return first, and database response can be issued caller by notification message.
Embodiment
At the automobile monitoring software application database based on the ARM platform, the positional information that 1000 automobile GPS of usage data library storage send.At first revise and show capacity setting in the configuration file, it is 10,000 that the table capacity for example is set.Database core is according to the table capacity application internal memory creation database table that sets.Secondly index type corresponding to option table stored GPS positional information HASH index speed because be.It is 1 second that the timing deposit time then is set, and can guarantee like this to only have second loss of data of level.It is comparatively abundant to have 512M in the native system, and therefore the rollback buffer zone being set is 100M, can protected data carry out rollback when guaranteeing a large amount of mistake to occur like this.Electrifying startup database after more than configuration is finished, database at first loads data into the internal storage data table from Flash, when having new location information data to upgrade, database root according to index navigate to that corresponding record inserts, deletion and retouching operation.As shown in Figure 3, memory database provides the interfaces such as insertion, deletion, modification, inquiry to application, and the user is by these interface operation databases.Interface can guarantee security and the data consistency of user's multi-course concurrency accessing database.Memory database is by regularly deposit mechanism, do not have second with the database data timed backup in the internal memory in hard disk, and guaranteeing can be because of the power down obliterated data.When database restarts, data are re-loaded in the memory database from hard disk simultaneously.
When promoting database work efficient, use memory database can directly save hardware investment and system maintenance cost.The actual deployment experience shows, for reaching the requirement of real-time of specific heavily loaded system, use memory database and do not reduce more than 2/3 than not using needed CPU quantity, and only need to amount of memory be improved on a small quantity, thereby greatly saved the funds fund input, system stability is improved on the contrary.
Claims (5)
1. embedded real-time internal memory database, comprise database core, database control and management module, three modules of database interface, it is characterized in that: database core defines and organizes data, database interface carries out access visit to database, and the Data Control administration module is controlled the state of database.
2. a kind of embedded real-time internal memory database according to claim 1, it is characterized in that: database core adopts relational data model, the tables of data of database core adopts the bivariate table structure, every bivariate table is used for creating major key index and the secondary index that can accelerate bivariate table record retrieval speed, and database control and management module is by the load-on module that powers on, regularly the module of depositing forms.
3. a kind of embedded real-time internal memory database according to claim 1 and 2 is characterized in that: described database control and management module is divided into the loading that powers on, two submodules of depositing regularly.
4. a kind of embedded real-time internal memory database according to claim 1 and 2, it is characterized in that: memory database adopts the mode of double-deck storage, be to have real-time data in the memory table, and the data in regularly will showing are kept in the physical hard disk, after the database power down, restart and from hard disk, to reload data.
5. a kind of embedded real-time internal memory database according to claim 2, it is characterized in that: every table all can create index, comprises that a major key index and quantity are greater than one secondary index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103342726A CN102880678A (en) | 2012-09-11 | 2012-09-11 | Embedded real-time memory database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103342726A CN102880678A (en) | 2012-09-11 | 2012-09-11 | Embedded real-time memory database |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102880678A true CN102880678A (en) | 2013-01-16 |
Family
ID=47482004
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012103342726A Pending CN102880678A (en) | 2012-09-11 | 2012-09-11 | Embedded real-time memory database |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102880678A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104657467A (en) * | 2015-02-11 | 2015-05-27 | 南京国电南自美卓控制系统有限公司 | Data pushing framework with subscription/release for real-time main memory database |
| CN104778125A (en) * | 2015-04-03 | 2015-07-15 | 无锡天脉聚源传媒科技有限公司 | Method and system for memory management |
| CN105447043A (en) * | 2014-09-01 | 2016-03-30 | 光宝科技股份有限公司 | Database and data access method thereof |
| CN106897325A (en) * | 2015-12-21 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of data load method and device |
| CN106919666A (en) * | 2017-02-17 | 2017-07-04 | 武汉逸居思宏科技有限公司 | A kind of Embedded Main Memory Database of object-oriented |
| CN107025223A (en) * | 2016-01-29 | 2017-08-08 | 华为技术有限公司 | A kind of buffer management method and server towards multi-tenant |
| CN107767140A (en) * | 2017-10-02 | 2018-03-06 | 头等尝网络餐饮管理科技（深圳）有限公司 | Method of payment, device, equipment and readable storage medium storing program for executing |
| CN110232057A (en) * | 2019-05-29 | 2019-09-13 | 掌阅科技股份有限公司 | Data rewind method, electronic equipment, storage medium |
| CN110888934A (en) * | 2019-11-12 | 2020-03-17 | 中核控制系统工程有限公司 | Scatter-based big data storage method |
| CN112035522A (en) * | 2020-07-16 | 2020-12-04 | 中科驭数（北京）科技有限公司 | Database data acquisition method and device |
-
2012
- 2012-09-11 CN CN2012103342726A patent/CN102880678A/en active Pending
Non-Patent Citations (3)
| Title |
|---|
| 刘震等: "嵌入式实时数据库技术研究", 《电子产品世界》 * |
| 张效尉等: "内存数据库故障恢复策略研究", 《科技广场》 * |
| 肖迎元等: "嵌入式实时内存数据库故障恢复技术", 《计算机科学》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105447043A (en) * | 2014-09-01 | 2016-03-30 | 光宝科技股份有限公司 | Database and data access method thereof |
| CN105447043B (en) * | 2014-09-01 | 2018-09-11 | 光宝科技股份有限公司 | Database and its data access method |
| CN104657467B (en) * | 2015-02-11 | 2017-09-05 | 南京国电南自维美德自动化有限公司 | A kind of data-pushing framework with subscription/publication of real-time internal memory database |
| CN104657467A (en) * | 2015-02-11 | 2015-05-27 | 南京国电南自美卓控制系统有限公司 | Data pushing framework with subscription/release for real-time main memory database |
| CN104778125A (en) * | 2015-04-03 | 2015-07-15 | 无锡天脉聚源传媒科技有限公司 | Method and system for memory management |
| CN104778125B (en) * | 2015-04-03 | 2017-09-15 | 无锡天脉聚源传媒科技有限公司 | A kind of EMS memory management process and system |
| CN106897325A (en) * | 2015-12-21 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of data load method and device |
| CN107025223B (en) * | 2016-01-29 | 2019-11-22 | 华为技术有限公司 | A kind of buffer management method and server towards multi-tenant |
| CN107025223A (en) * | 2016-01-29 | 2017-08-08 | 华为技术有限公司 | A kind of buffer management method and server towards multi-tenant |
| CN106919666A (en) * | 2017-02-17 | 2017-07-04 | 武汉逸居思宏科技有限公司 | A kind of Embedded Main Memory Database of object-oriented |
| CN107767140A (en) * | 2017-10-02 | 2018-03-06 | 头等尝网络餐饮管理科技（深圳）有限公司 | Method of payment, device, equipment and readable storage medium storing program for executing |
| CN110232057A (en) * | 2019-05-29 | 2019-09-13 | 掌阅科技股份有限公司 | Data rewind method, electronic equipment, storage medium |
| CN110232057B (en) * | 2019-05-29 | 2021-03-12 | 掌阅科技股份有限公司 | Data rollback method, electronic device and storage medium |
| CN110888934A (en) * | 2019-11-12 | 2020-03-17 | 中核控制系统工程有限公司 | Scatter-based big data storage method |
| CN112035522A (en) * | 2020-07-16 | 2020-12-04 | 中科驭数（北京）科技有限公司 | Database data acquisition method and device |
| CN112035522B (en) * | 2020-07-16 | 2021-09-07 | 中科驭数（北京）科技有限公司 | Database data acquisition method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102880678A (en) | Embedded real-time memory database | |
| US10296615B2 (en) | In-memory database system | |
| US11119997B2 (en) | Lock-free hash indexing | |
| CN105630865B (en) | N-bit compressed versioned column data array for memory columnar storage | |
| US8346722B2 (en) | Replica placement strategy for distributed data persistence | |
| CN105630860B (en) | Database system with transaction control block index | |
| US6411964B1 (en) | Methods for in-place online reorganization of a database | |
| US6351744B1 (en) | Multi-processor system for database management | |
| CN101986649B (en) | Shared data center used in telecommunication industry billing system | |
| CN107766402A (en) | A kind of building dictionary cloud source of houses big data platform | |
| CN102054034A (en) | Implementation method for business basic data persistence of enterprise information system | |
| CN102254021A (en) | Method for constructing database based on virtual machine management system | |
| Sowell et al. | Minuet: A scalable distributed multiversion B-tree | |
| CN100424699C (en) | Attribute extensible object file system | |
| CN104021145A (en) | Mixed service concurrent access method and device | |
| CN102955792A (en) | Method for implementing transaction processing for real-time full-text search engine | |
| Cubukcu et al. | Citus: Distributed postgresql for data-intensive applications | |
| CN102163232A (en) | SQL (Structured Query Language) interface implementing method supporting IEC61850 object query | |
| CN110096509A (en) | Realize that historical data draws the system and method for storage of linked list modeling processing under big data environment | |
| CN103488710A (en) | Efficient-storage unsteady data structure for big data pages | |
| CN110175206A (en) | Intellectual analysis operational approach, system and medium for multiple database separation | |
| Mukherjee | Synthesis of non-replicated dynamic fragment allocation algorithm in distributed database systems | |
| CN109918450A (en) | Based on the distributed parallel database and storage method under analysis classes scene | |
| CN105095255A (en) | Data index creating method and device | |
| CN113821573A (en) | Mass data rapid retrieval service construction method, system, terminal and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130116 |