CN106909556A - The storage equalization methods and device of main memory cluster - Google Patents
The storage equalization methods and device of main memory cluster Download PDFInfo
- Publication number
- CN106909556A CN106909556A CN201510976653.8A CN201510976653A CN106909556A CN 106909556 A CN106909556 A CN 106909556A CN 201510976653 A CN201510976653 A CN 201510976653A CN 106909556 A CN106909556 A CN 106909556A
- Authority
- CN
- China
- Prior art keywords
- key
- original
- data
- level key
- burst number
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2228—Indexing structures
- G06F16/2255—Hash tables
-
- 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
- G06F16/22—Indexing; Data structures therefor; Storage structures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses the storage equalization methods and device of a kind of main memory cluster, it is related to database field.Method therein includes:Cutting is carried out to the table sectional of original one-level key K1 and obtains burst number;According to original one-level key K1, the balanced one-level key K1 ' of burst number generation;Key assignments data are stored using the data storage format of (K1 ', DO);Wherein, original one-level key K1 is the primitive logic key of customized objects DO, and the original one-level key K1 is made up of database segment, data table segment, table sectional, and balanced one-level key K1 ' is the logic of Equilibrium key of customized objects DO.So as to realize data being uniformly distributed on main memory cluster back end, and then make full use of the memory capacity of main memory cluster.
Description
Technical field
The present invention relates to database field, the storage equalization methods of more particularly to a kind of main memory cluster
And device.
Background technology
Uniformity Hash is done to the key (key) in cluster by proxy server, ideally can
Ensure key being uniformly distributed on clustered node.Because the data volume of subregion is generally different, point
It is unbalanced that the difference of area's data volume can cause the storage resource of clustered node to take.
As shown in Figure 1, it is assumed that cluster has 4 nodes, and the storage of each node is 10.It is divided to two
8 partition datas of secondary storage, store 4 subregions (4 keys), data volume difference for the first time
It is 1,2,4,8;Store 4 subregions (4 keys) again for the second time, data volume is all 2.
If the partition data of storage twice is all uniformly stored on 4 nodes, then 4 sections
The data volume of point is respectively 3,4,6,10.If inserting data to clustered node again, very may be used
Energy data are hashing onto on the node for being stored as 10 by uniformity, then storage failure.Now, other
3 nodes have different degrees of slack resources but unavailable.
It can be seen that, traditional uniformity hash algorithm does not consider that subregion influences, directly to data member
Peer processes, while the mode of addition dummy node exacerbates the feelings of subregion skewness in recent years
Condition, it is therefore necessary to solve partition data concentrations and cause overall performance in some service nodes
Bottleneck problem.
The content of the invention
The technical problems to be solved by the invention are:How to realize data in main memory cluster back end
On be uniformly distributed, and then make full use of the memory capacity of main memory cluster.
One side according to embodiments of the present invention, there is provided a kind of storage of main memory cluster is balanced
Method, including:Cutting is carried out to the table sectional of original one-level key K1 and obtains burst number;Root
According to original one-level key K1, the balanced one-level key K1 ' of burst number generation;Using (K1 ', DO)
Data storage format stores key assignments data;Wherein, original one-level key K1 is customized objects DO
Primitive logic key, the original one-level key K1 is by database segment, data table segment, table subregion
Duan Zucheng, balanced one-level key K1 ' are the logic of Equilibrium key of customized objects DO.
Other side according to embodiments of the present invention, there is provided a kind of storage of main memory cluster is balanced
Device, including:Cutting module, cutting is carried out for the table sectional to original one-level key K1
Obtain burst number;Balanced one-level key generation module, for according to original one-level key K1, burst number
The balanced one-level key K1 ' of generation;Key assignments data memory module, for using (K1 ', DO)
Data storage format stores key assignments data;Wherein, original one-level key K1 is customized objects DO
Primitive logic key, the original one-level key K1 is by database segment, data table segment, table subregion
Duan Zucheng, balanced one-level key K1 ' are the logic of Equilibrium key of customized objects DO.
The present invention at least has advantages below:
Burst number is obtained by the further cutting of table sectional to original key, then according to original
Key and burst number rebuild balanced key, and the storage of data is carried out based on balanced key, realize number
According to being uniformly distributed on main memory cluster back end, and then make full use of the storage of main memory cluster
Capacity.
By referring to the drawings to the detailed description of exemplary embodiment of the invention, the present invention
Further feature and its advantage will be made apparent from.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will
The accompanying drawing to be used needed for embodiment or description of the prior art is briefly described, it is clear that
Ground, drawings in the following description are only some embodiments of the present invention, for the common skill in this area
For art personnel, without having to pay creative labor, can also be obtained according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 shows that traditional hash algorithm causes the storage resource of clustered node to take unbalanced showing
It is intended to.
Fig. 2 shows that the flow of one embodiment of the storage equalization methods of main memory cluster of the present invention is shown
It is intended to.
Fig. 3 shows that the present invention carries out cutting and obtains burst to the table sectional of original one-level key K1
Number schematic diagram.
Fig. 4 shows that the present invention carries out cutting and obtains burst to the table sectional of original one-level key K1
Number one embodiment schematic diagram.
Fig. 5 shows that the present invention carries out cutting and obtains burst number to the table sectional of original one-level key K1
One embodiment schematic diagram.
Fig. 6 shows the schematic diagram of one embodiment of the storage balancer of main memory cluster of the present invention.
Fig. 7 shows the schematic diagram of one embodiment of key assignments data memory module of the present invention.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, to the technical scheme in the embodiment of the present invention
It is clearly and completely described, it is clear that described embodiment is only a real part of the invention
Example is applied, rather than whole embodiments.Below to the description reality of at least one exemplary embodiment
It is merely illustrative on border, never as to the present invention and its application or any limitation for using.
Based on the embodiment in the present invention, those of ordinary skill in the art are not before creative work is made
The every other embodiment for being obtained is put, the scope of protection of the invention is belonged to.
The storage equalization methods of the main memory cluster of one embodiment of the invention are described with reference to Fig. 2.
Fig. 2 shows that the flow of one embodiment of the storage equalization methods of main memory cluster of the present invention is shown
It is intended to.As shown in Fig. 2 the method for the embodiment includes:
Step S202, carries out cutting and obtains burst number to the table sectional of original one-level key K1.
Original one-level key K1 is the primitive logic key of customized objects DO, and original one-level key K1 is by data
Libraries section, data table segment, table sectional composition, so as to provide database, data on Redis clusters
The Method of Data Organization of table, subregion.Wherein, database db is maximum data isolation unit, is
One logical concept.User can create multiple databases.In general, the data of a class should put
In a memory database.Tables of data table is a class theme, business, concept identical data
Set, such as user's table, detail list, are a logical concepts.Table subregion partition is to be based on
Redis cluster features, the unit that the further logic of big data table is split.Data in different subregions
It is to be stored separately, makes the distribution of data more uniform.Additionally, database, tables of data, subregion this
A little logical concepts can be described with metadata.It is STORE, value for storing the key name of metadata
Type is Hash (Hash table).All metadata informations, are all stored in STORE as key
Under Hash table.The metadata description information of all databases is all stored in the form of Key-Value
In metadatabase (Hash table), the key name of database can be DB_ [database name].All data
The metadata description information of table be all stored in metadatabase in the form of the Key-Value in (Hash table).
The key name of tables of data can be TB_ [database name] _ [data table name], so as to embody number on key name
According to table and the relation of database.Data partition is the further logical partitioning of tables of data, data partition
In the metadata of the tables of data preserved in the way of metadata.The key name of data partition can be [data
Library name] _ [data table name] _ [zone name].
Wherein, (K2, V) is stored in customized objects DO, weak linkage K2 is customized objects
The data key of DO, V is the data value of customized objects DO.
Step S204, according to original one-level key K1, the balanced one-level key K1 ' of burst number generation.
Wherein, balanced one-level key K1 ' is the logic of Equilibrium key of customized objects DO.
Step S206, key assignments data are stored using the data storage format of (K1 ', DO).
By the above method, the table sectional to original one-level key K1 is further segmented,
Then balanced key is rebuild according to original key and burst number, and data is carried out based on balanced key
Storage, realizes data being uniformly distributed on main memory cluster back end, and then make full use of
The memory capacity of main memory cluster.
Fig. 3 shows that the present invention carries out cutting and obtains burst to the table sectional of original one-level key K1
Number schematic diagram.Burst is that the nodal point number stored by user is calculated.During non-data storage,
Not can determine that a subregion is specifically divided into how many bursts.When user stores Key-Value,
Cryptographic Hash, then burst number remainder according to plan are taken to Key.The result for calculating is complete with former subregion
Whole key name combination, forms new burst key name,
The table to original one-level key K1 point of one embodiment of the invention is described with reference to Fig. 4
Section carries out cutting and obtains burst method.
Fig. 4 shows that the present invention carries out cutting and obtains burst to the table sectional of original one-level key K1
Number one embodiment schematic diagram.As shown in figure 4, stored in customized objects DO (K2,
V), weak linkage K2 is the data key of customized objects DO, and V is the number of customized objects DO
According to value.The table sectional of original one-level key K1 is carried out cutting obtain burst number one kind it is specific
Implementation method includes:
Step S402, Hash operation is carried out to original one-level key K1 and weak linkage K2.
Step S404, remainder fortune is carried out by Hash operation gained cryptographic Hash to plan burst number M
Calculate.
Step S406, using the result of complementation as burst number.
For example, clustered node number is 4, the maximum burst number of subregion is designed as 40.So user
To one new value of storage in the table1 of db1:Key-Value, the corresponding subregion of the value is:
String slice=String.valueOf (Math.abs (Key.hashCode ())/40).Assuming that meter
The result of calculation is 23.If complete subregion key name is db1_table1_p1, new subregion
Entitled db1_table1_p1_23.
Data " cutting " are further divided into piece on the basis of subregion.When number of slices is more than nodes
When, section will be evenly distributed on different nodes according to the characteristic of TwemProxy.It is former
Number of slices is then gone up more, data storage granularity is smaller, and data distribution is more uniform, so as to realize number
According to being uniformly distributed on main memory cluster back end, and then the storage of main memory cluster is made full use of to hold
Amount.But slice of data also need not be excessive, N (N≤10) times of nodes is generally kept in i.e.
Can.
A table to original one-level key K1 of another embodiment of the invention is described with reference to Fig. 5
Sectional carries out cutting and obtains burst method.
Fig. 5 shows that the present invention carries out cutting and obtains burst to the table sectional of original one-level key K1
Number another embodiment schematic diagram.As shown in figure 5, to the table point of original one-level key K1
Section carries out cutting and obtains another concrete methods of realizing of burst number including:
Step S502, md5 computings are carried out to original one-level key K1 and weak linkage K2;
Step S504, takes rear n of md5 computings gained digest value as burst number, n institute
The greatest measure that can be represented is not more than plan burst number M.
Burst number is obtained by carrying out md5 computings to original one-level key K1 and weak linkage K2,
Can equally cause that data storage granularity is smaller, data distribution is more uniform, so as to realize that data exist
Being uniformly distributed on main memory cluster back end, and then make full use of the memory capacity of main memory cluster.
Compared with the embodiment shown in Fig. 4, md5 algorithms can be according to original one-level key K1 and weak linkage
K2 calculates the burst number that result is unique value, but md5 algorithms are higher to the occupancy of CPU,
Performance can be influenceed under big concurrent pressure.
Additionally, customized objects DO can be stored by the form of MAP.
The storage balancer of the main memory cluster of one embodiment of the invention is described with reference to Fig. 6.
Fig. 6 shows the schematic diagram of one embodiment of the storage balancer of main memory cluster of the present invention.
As shown in fig. 6, the storage balancer 60 of the main memory cluster of the embodiment includes:
Cutting module 602, carries out cutting and is divided for the table sectional to original one-level key K1
Piece number.
Balanced one-level key generation module 604, for being generated according to original one-level key K1, burst number
Balanced one-level key K1 '.
Key assignments data memory module 606, for the data storage format using (K1 ', DO)
Storage key assignments data.
Wherein, original one-level key K1 is the primitive logic key of customized objects DO, described original
One-level key K1 is made up of database segment, data table segment, table sectional, balanced one-level key K1 '
It is the logic of Equilibrium key of customized objects DO.
In one embodiment, (K2, V) is stored in customized objects DO, weak linkage K2
It is the data key of customized objects DO, V is the data value of customized objects DO;
Cutting module 602 is used for:Hash operation is carried out to original one-level key K1 and weak linkage K2;
Hash operation gained cryptographic Hash is carried out into complementation to plan burst number M;By complementation
Result is used as burst number.
In another embodiment, cutting module 602 can be used for:To original one-level key K1
Md5 computings are carried out with weak linkage K2;Take rear n conduct point of md5 computings gained digest value
Piece number, the n greatest measure that can be represented is not more than plan burst number M.
The storage balancer 60 of main memory cluster can also include:
Plan burst number determining module 608, described in the nodes for being stored according to user determine
Plan burst number M, the plan burst number M are N times of nodes, and wherein N is little
In 10 natural number.
The key assignments data memory module of one embodiment of the invention is described with reference to Fig. 7.
Fig. 7 shows the schematic diagram of one embodiment of key assignments data memory module of the present invention.Such as Fig. 7
Shown, the key assignments data memory module 606 of the embodiment includes customized objects memory cell 6062,
For storing customized objects DO in the form of MAP.
One of ordinary skill in the art will appreciate that realizing that all or part of step of above-described embodiment can
To be completed by hardware, it is also possible to instruct the hardware of correlation to complete by program, described journey
Sequence can be stored in a kind of computer-readable recording medium, and storage medium mentioned above can be
Read-only storage, disk or CD etc..
Presently preferred embodiments of the present invention is the foregoing is only, is not intended to limit the invention, it is all at this
Within the spirit and principle of invention, any modification, equivalent substitution and improvements made etc. all should be wrapped
It is contained within protection scope of the present invention.
Claims (10)
1. storage equalization methods of a kind of main memory cluster, including:
Cutting is carried out to the table sectional of original one-level key K1 and obtains burst number;
According to original one-level key K1, the balanced one-level key K1 ' of burst number generation;
Key assignments data are stored using the data storage format of (K1 ', DO);
Wherein, original one-level key K1 is the primitive logic key of customized objects DO, described original
One-level key K1 is made up of database segment, data table segment, table sectional, balanced one-level key K1 '
It is the logic of Equilibrium key of customized objects DO.
2. method according to claim 1, it is characterised in that wherein, customized objects
(K2, V) is stored in DO, weak linkage K2 is the data key of customized objects DO, and V is fixed
The data value of object DO processed;
The table sectional to original one-level key K1 carries out cutting and obtains burst number including:
Hash operation is carried out to original one-level key K1 and weak linkage K2;
Hash operation gained cryptographic Hash is carried out into complementation to plan burst number M;
Using the result of complementation as burst number.
3. method according to claim 1, it is characterised in that wherein, customized objects
(K2, V) is stored in DO, weak linkage K2 is the data key of customized objects DO, and V is fixed
The data value of object DO processed;
The table sectional to original one-level key K1 carries out cutting and obtains burst number including:
Md5 computings are carried out to original one-level key K1 and weak linkage K2;
Rear n of md5 computings gained digest value is taken as burst number, n can represent most
Big numerical value is not more than plan burst number M.
4. according to the method in claim 2 or 3, it is characterised in that also include:
Determine plan the burst number M, the plan burst number according to the nodes that user stores
M is N times of nodes, and wherein N is no more than 10 natural number.
5. method according to claim 1, it is characterised in that wherein, the customization is right
As DO is stored in the form of MAP.
6. the storage balancer of a kind of main memory cluster, including:
Cutting module, carries out cutting and obtains burst for the table sectional to original one-level key K1
Number;
Balanced one-level key generation module, for balanced according to original one-level key K1, burst number generation
One-level key K1 ';
Key assignments data memory module, stores for the data storage format using (K1 ', DO)
Key assignments data;
Wherein, original one-level key K1 is the primitive logic key of customized objects DO, described original
One-level key K1 is made up of database segment, data table segment, table sectional, balanced one-level key K1 '
It is the logic of Equilibrium key of customized objects DO.
7. device according to claim 6, it is characterised in that wherein, customized objects
(K2, V) is stored in DO, weak linkage K2 is the data key of customized objects DO, and V is fixed
The data value of object DO processed;
The cutting module is used for:
Hash operation is carried out to original one-level key K1 and weak linkage K2;
Hash operation gained cryptographic Hash is carried out into complementation to plan burst number M;
Using the result of complementation as burst number.
8. device according to claim 6, it is characterised in that wherein, customized objects
(K2, V) is stored in DO, weak linkage K2 is the data key of customized objects DO, and V is fixed
The data value of object DO processed;
The cutting module is used for:
Md5 computings are carried out to original one-level key K1 and weak linkage K2;
Rear n of md5 computings gained digest value is taken as burst number, n can represent most
Big numerical value is not more than plan burst number M.
9. the device according to claim 7 or 8, it is characterised in that also include:
Plan burst number determining module, the nodes for being stored according to user determine the plan
Burst number M, the plan burst number M are N times of nodes, and wherein N is no more than 10
Natural number.
10. device according to claim 6, it is characterised in that key assignments data storage mould
Block also includes customized objects memory cell, for storing customized objects DO in the form of MAP.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510976653.8A CN106909556B (en) | 2015-12-23 | 2015-12-23 | Memory cluster storage balancing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510976653.8A CN106909556B (en) | 2015-12-23 | 2015-12-23 | Memory cluster storage balancing method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106909556A true CN106909556A (en) | 2017-06-30 |
CN106909556B CN106909556B (en) | 2020-03-20 |
Family
ID=59200301
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510976653.8A Active CN106909556B (en) | 2015-12-23 | 2015-12-23 | Memory cluster storage balancing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106909556B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109101635A (en) * | 2018-08-16 | 2018-12-28 | 广州小鹏汽车科技有限公司 | A kind of data processing method and device based on Redis Hash structure |
CN109344161A (en) * | 2018-12-04 | 2019-02-15 | 大唐网络有限公司 | A kind of mass data storage means based on mongodb |
CN110287197A (en) * | 2019-06-28 | 2019-09-27 | 微梦创科网络科技(中国)有限公司 | A kind of date storage method, moving method and device |
WO2020259309A1 (en) * | 2019-06-28 | 2020-12-30 | 苏宁云计算有限公司 | Multi-dimension data query method and apparatus |
CN113791740A (en) * | 2021-11-10 | 2021-12-14 | 深圳市杉岩数据技术有限公司 | Method for recording object storage bucket statistics and counting |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101876983A (en) * | 2009-04-30 | 2010-11-03 | 国际商业机器公司 | Method for partitioning database and system thereof |
CN103699676A (en) * | 2013-12-30 | 2014-04-02 | 厦门市美亚柏科信息股份有限公司 | MSSQL SERVER based table partition and automatic maintenance method and system |
CN103838770A (en) * | 2012-11-26 | 2014-06-04 | 中国移动通信集团北京有限公司 | Logic data partition method and system |
US20140330801A1 (en) * | 2013-05-06 | 2014-11-06 | International Business Machines Corporation | Lock-free creation of hash tables in parallel |
US20140351239A1 (en) * | 2013-05-23 | 2014-11-27 | Microsoft Corporation | Hardware acceleration for query operators |
-
2015
- 2015-12-23 CN CN201510976653.8A patent/CN106909556B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101876983A (en) * | 2009-04-30 | 2010-11-03 | 国际商业机器公司 | Method for partitioning database and system thereof |
CN103838770A (en) * | 2012-11-26 | 2014-06-04 | 中国移动通信集团北京有限公司 | Logic data partition method and system |
US20140330801A1 (en) * | 2013-05-06 | 2014-11-06 | International Business Machines Corporation | Lock-free creation of hash tables in parallel |
US20140351239A1 (en) * | 2013-05-23 | 2014-11-27 | Microsoft Corporation | Hardware acceleration for query operators |
CN103699676A (en) * | 2013-12-30 | 2014-04-02 | 厦门市美亚柏科信息股份有限公司 | MSSQL SERVER based table partition and automatic maintenance method and system |
Non-Patent Citations (1)
Title |
---|
刘超 等: "Oracle数据库系统优化调整", 《信息安全与技术》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109101635A (en) * | 2018-08-16 | 2018-12-28 | 广州小鹏汽车科技有限公司 | A kind of data processing method and device based on Redis Hash structure |
CN109101635B (en) * | 2018-08-16 | 2020-09-11 | 广州小鹏汽车科技有限公司 | Data processing method and device based on Redis Hash structure |
CN109344161A (en) * | 2018-12-04 | 2019-02-15 | 大唐网络有限公司 | A kind of mass data storage means based on mongodb |
CN110287197A (en) * | 2019-06-28 | 2019-09-27 | 微梦创科网络科技(中国)有限公司 | A kind of date storage method, moving method and device |
WO2020259309A1 (en) * | 2019-06-28 | 2020-12-30 | 苏宁云计算有限公司 | Multi-dimension data query method and apparatus |
CN113791740A (en) * | 2021-11-10 | 2021-12-14 | 深圳市杉岩数据技术有限公司 | Method for recording object storage bucket statistics and counting |
CN113791740B (en) * | 2021-11-10 | 2022-02-18 | 深圳市杉岩数据技术有限公司 | Method for recording object storage bucket statistics and counting |
Also Published As
Publication number | Publication date |
---|---|
CN106909556B (en) | 2020-03-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20200356901A1 (en) | Target variable distribution-based acceptance of machine learning test data sets | |
CN107040582B (en) | Data processing method and device | |
Demertzis et al. | Searchable encryption with optimal locality: Achieving sublogarithmic read efficiency | |
US10432639B1 (en) | Security management for graph analytics | |
US11175995B2 (en) | Data protection for a document database system | |
US20160299919A1 (en) | Management of Intermediate Data Spills during the Shuffle Phase of a Map-Reduce Job | |
CN106909556A (en) | The storage equalization methods and device of main memory cluster | |
US9613037B2 (en) | Resource allocation for migration within a multi-tiered system | |
CN110347651B (en) | Cloud storage-based data synchronization method, device, equipment and storage medium | |
US20140122706A1 (en) | Method for determining system topology graph changes in a distributed computing system | |
US10331669B2 (en) | Fast query processing in columnar databases with GPUs | |
US20190140902A1 (en) | Centralized configuration data in a distributed file system | |
Patwary et al. | Window-based streaming graph partitioning algorithm | |
US11270227B2 (en) | Method for managing a machine learning model | |
Peng et al. | A GPU‐based approach for massive model rendering with frame‐to‐frame coherence | |
CN115048254B (en) | Simulation test method, system, equipment and readable medium for data distribution strategy | |
US7792966B2 (en) | Zone control weights | |
CN106909557B (en) | Memory cluster storage method and device and memory cluster reading method and device | |
US9703788B1 (en) | Distributed metadata in a high performance computing environment | |
US20160357795A1 (en) | Method and apparatus for data mining | |
CN110928941B (en) | Data fragment extraction method and device | |
CN116578641A (en) | Database separation method and system based on ketama algorithm | |
KR20210137612A (en) | Device, method, system and computer readable storage medium for managing blockchain | |
KR101772955B1 (en) | Record processing method using index data structure in distributed processing system based on mapreduce | |
Jain et al. | Bloom Filter in Cloud Storage for Efficient Data Membership Identification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220128 Address after: 100007 room 205-32, floor 2, building 2, No. 1 and No. 3, qinglonghutong a, Dongcheng District, Beijing Patentee after: Tianyiyun Technology Co.,Ltd. Address before: No.31, Financial Street, Xicheng District, Beijing, 100033 Patentee before: CHINA TELECOM Corp.,Ltd. |
|
TR01 | Transfer of patent right |