CN104281717B - A kind of method for setting up magnanimity ID mapping relations - Google Patents
A kind of method for setting up magnanimity ID mapping relations Download PDFInfo
- Publication number
- CN104281717B CN104281717B CN201410603292.8A CN201410603292A CN104281717B CN 104281717 B CN104281717 B CN 104281717B CN 201410603292 A CN201410603292 A CN 201410603292A CN 104281717 B CN104281717 B CN 104281717B
- Authority
- CN
- China
- Prior art keywords
- mapping relations
- index
- identification code
- mapping
- major key
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/957—Browsing optimisation, e.g. caching or content distillation
- G06F16/9574—Browsing optimisation, e.g. caching or content distillation of access to content, e.g. by caching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/955—Retrieval from the web using information identifiers, e.g. uniform resource locators [URL]
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Present invention is disclosed a kind of method for setting up magnanimity ID mapping relations, comprise the following steps：(1) the major key identification code pressed in mapping relations A selects corresponding block；(2) judge whether current block there are the mapping relations, if so, step (3) is then performed, if it is not, then performing step (5)；(3) the mapping relations B stored before the major key identification code is read；(4) judge mapping relations B whether containment mapping relation A, if it is not, then perform step (5), if so, then perform step (7)；(5) merge mapping relations A and B, and internal memory is updated to the mapping relations after one group of new mapping relations or merging and index entirely；(6) will mapping data write-in disk；And (7) insertion mapping relations success.The present invention can realize that unit supports up to 1B groups mapping relations, expand storage capacity using multistage composite index and lift query performance, mapping relations are recorded by disk index partition, index mapping relations where realization is inquired about by any ID entirely by internal memory.
Description
Technical field
The present invention relates to web advertisement field, more particularly to a kind of method for setting up magnanimity ID mapping relations.
Background technology
It is well known that Internet advertising is the topmost profit model of internet industry, flow, which is cashed, turns into internet business
The very important evaluation criterion of industry product.With the development of internet crowd's directional technology, Internet advertising also begins to break away from list
First, old-fashioned and inflexible trade mode, then change to more accurate efficient trade mode.This middle real time bid (Real Time
Bidding, abbreviation RTB) play an important ring.
In the bidding of RTB, the visit that can be typically provided in bid request containing an ADX (Ad Exchange platforms)
Visitor's mark, this mark can be understood as the cookie similar to USERID, but can never be in Ad Exchange systems
The ID in portion, can be typically using non-reversible AES a hash again to DSP (Demand-Side Platform abbreviation,
The purchaser of flow can be regarded as, be advertiser's service), USERID after encryption we be called USERID '.And DSP mono-
As need the various dimension datas that are directed in bidrequest, including PV information, user's characteristic information, ad spot information etc. is determined
Whether current exposure is bought, " marketing again (the retargeting) " technology for going back popular has to rely on user's mark, so this
Individual USERID ' is DSP needs, and DSP needs voluntarily to safeguard a USERID ' mapping relations, is exactly the USERID ' and oneself
One mapping of user's mark of definition.
Typically to provide key-value inquiry services can use NoSql technologies to be solved, but with the ADX of access
Platform and advertiser are more and more, and the amount for the ID mapping relations that DSP side is safeguarded also gradually increases, the conventional NoSql based on internal memory
Technology (memcached, redis etc.) is limited by memory size, while there is larger expansion rate to key-value pair, unit can not be supported
The up to storage and inquiry of 1B groups ID mapping relations.Many based on the small clusters of internal memory NoSql are compared machine resources waste, nothing
Method is concentrated use in, and there are problems that.It is other based on disk file NoSql technologies (LevelDB,
SSDB etc.) substantial amounts of random challenge request can not be met.For this phenomenon, the storages of urgent need to resolve magnanimity ID mapping relations with
And corresponding frequent inquiry request is provided.
In view of this, a kind of device of the storage of magnanimity ID mapping relations and inquiry arises at the historic moment.
The content of the invention
The invention provides a kind of method for setting up magnanimity ID mapping relations, the difficulty of prior art, Neng Goushi are overcome
Existing unit supports up to 1B (1,000,000,000) group mapping relations, expands storage capacity using multistage composite index and lifts query performance,
Mapping relations are recorded by disk index partition, mapping relations where realization is inquired about by any ID are indexed entirely by internal memory.
The invention provides a kind of method for setting up magnanimity ID mapping relations, comprise the following steps：
(1) the major key identification code pressed in mapping relations A selects corresponding block；
(2) judge whether current block there are the mapping relations, if so, step (3) is then performed, if it is not, then performing step
(5)；
(3) the mapping relations B stored before the major key identification code is read；
(4) judge mapping relations B whether containment mapping relation A, if it is not, then perform step (5), if so, then performing step
(7)；
(5) merge mapping relations A and B, and the mapping relations after one group of new mapping relations or merging are updated
Internal memory is indexed entirely；
(6) will mapping data write-in disk；And
(7) insertion mapping relations success,
Wherein, index is the index key of all mapping relations corresponding thereto to internal memory described in step (5) entirely.
Preferably, if having pointed to other major key identification code in index, it is covered using current major key identification code,
This group of mapping relations are written in disk again.
The method for setting up magnanimity ID mapping relations of the present invention includes distributing unique by different types for all ID
Identification code, and mapping relations by identification code be stored in internal memory full index in；The identification code choosing of major key in mapping relations
Corresponding subregion is selected to be stored；When major key is indexed and solid entirely in the presence of, it is necessary to be selectively updated internal memory by identification code
State storage device.Support up to 1B (1,000,000,000) to organize mapping relations by above method unit of the present invention, indexed using multistage composite
Expand storage capacity and lift query performance, mapping relations are recorded by disk index partition, index realization entirely by internal memory logical
Cross mapping relations where any ID inquiries.The present invention has first party Cookie and third party Cookie mapping library so that DSP
It can effectively be bidded according to Users'Data Analysis, realize the maximization of interests.Further DMP platforms (Data
Management Platform.DMP stores flow, the various features information of audient) just it is able to the scattered the first, the 3rd
Number formulary realizes the unified identification of audient according to being integrated.
The present invention is further illustrated below in conjunction with drawings and Examples.
Brief description of the drawings
Fig. 1 is the flow chart of the method for setting up magnanimity ID mapping relations of the present invention.
Embodiment
A kind of specific embodiment of the present invention is introduced below by Fig. 1.
Embodiment 1
As shown in figure 1, a kind of method for setting up magnanimity ID mapping relations of the present invention, comprises the following steps：
(1) the major key identification code pressed in mapping relations A selects corresponding block；
(2) judge whether current block there are the mapping relations, if so, step (3) is then performed, if it is not, then performing step
(5)；
(3) the mapping relations B stored before the major key identification code is read；
(4) judge mapping relations B whether containment mapping relation A, if it is not, then perform step (5), if so, then performing step
(7)；
(5) merge mapping relations A and B, and the mapping relations after one group of new mapping relations or merging are updated
Internal memory is indexed entirely；
(6) will mapping data write-in disk；And
(7) insertion mapping relations success,
Wherein, index is the index key of all mapping relations corresponding thereto to internal memory described in step (5) entirely.
If index in pointed to other major key identification code, cover it using current major key identification code, then by this
Group mapping relations are written in disk.
The embodiment of the present invention is as follows：
The basic memory cell of mapping relations in the present invention
One group of mapping relations as an independent memory cell, including：
By the DSP side ID correspondence major keys in above-mentioned background knowledge.
ADX side and the ID manipulative indexing keys of advertiser first party audient, they are all each to be used for according to certain algorithm
One user of unique mark.But do not ensure to be not in conflict during cross-platform or advertiser, so also needing to Record ID source, i.e.,
The type of platform (or advertiser).
Wherein index key comprises at least one, so every group of ID mapping relations may be with following form tissue.
Major key：{ index key source 1：Index key 1, index key source 2：Index key 2...... }.
Identification code is uniquely distributed for all key assignments, is not in conflicting mark in same type of key assignments
Code, identification code is mainly used in setting up different indexes, is indicated using 64 integers, internal memory can be saved to greatest extent.
Solid magnetic disc is indexed
Carried out during data Cun Chudao disks by according to major key between subregion (Block) index, the storage location of each subregion
It is separate and continuous, the insertion change of record mapping relations and all records deleted inside each Block, these
Operation is all to be recorded in the way of additional write-in (APPEND) inside Block.
In same subregion, the present invention is established in major key identification code and data in magnetic disk one by one by the way of forward direction index
Then mapping relations, can read corresponding mapping relations according to the storage location in the quick positioning disk of major key.
Internal memory is indexed entirely
The present invention also provides the function that inquiry mapping relations are carried out according to index key simultaneously, can be by this by step conversion
Class inquiry request is converted into inquiring about mapping relations according to major key.The present invention is by the way of multiple cross index, and every group of mapping is closed
The separate sources of meeting recording indexes key, has used HASH indexes and the identification code of BTree index record index keys to be mapped in system
Major key identification code, obtain after major key identification code reading corresponding mapping relations further according to above-mentioned disk index.
For the insertion of one group of mapping relations, corresponding Block, such as major key can be found out according to major key identification code first
Identification code is already present on the Block, then reads the mapping relations preserved before, is contrasted with the progress that is currently inserted into.
If new mapping result is contained in the mapping relations being inserted into, or mapping result is changed, then it is right
Two groups of mapping relations merge (Merge), are defined if any result of the conflict then in the mapping relations being inserted into.
Internal memory is updated for the mapping relations after one group of new mapping relations or merging to index entirely, if in index
Other major key identification code has been pointed to, then has covered it using current major key identification code.This group of mapping relations are written to magnetic again
In disk.
Inquiry
, can be according to the full search index of internal memory to correspondence if inquiry request is not major key for the inquiry of mapping relations
Major key identification code, such as can not find, then directly return Query Result for sky.
So inquiry request finally may be converted into inquiring about mapping relations according to major key identification code.
Likewise, it is similar with insertion request, corresponding Block can be found out according to major key identification code first, if major key
Identification code is not already present in the Block, then it is sky to return to Query Result, and otherwise reading disk file takes out mapping relations and returned
Return.
Delete
One group of mapping relations is deleted, if removal request is not major key identification code, can be arrived according to the full search index of internal memory
Corresponding major key identification code, such as can not find, then directly returns.
So removal request finally may be converted into deleting mapping relations according to major key identification code.
Likewise, corresponding Block can be found out according to major key identification code first, disk index is deleted, if primary key
Code is known not already present in the Block, then can not delete direct return.
In order to lift the write performance of disk, all operations are all in the way of Append, so correspondence one deletes behaviour
Make, what is write in disk is that the mapping relations for there was only major key identification code (are different from least one index under normal circumstances
The mapping relations of key).
Need to delete internal memory while deleting disk index and index entirely, by major key identification code first by mapping relations from disk
In read out, further according in result index key identification code delete internal memory full index in respective record.
It is noted that whether current index key identification code still points to the master when deletion internal memory is indexed entirely in addition
Key identification code, if not, representing that the index key has been recorded in another group of mapping relations, then this index need not be deleted
Remove.
Defragmentation
The change and deletion of mapping relations can be caused to the data in disk file " cavity ", long-term accumulated, which is got off, to be caused
The waste of memory space is, it is necessary to carry out disk fragmentses arrangement.
This method realizes a kind of de-fragmentation strategies of highly effective and safe based on disk index to be included
(1) snapshot is set up to the subregion for needing to carry out defragmentation；
(2) exchange area is set up according to snapshot；
(3) the mapping relations data of former subregion are write in the disk file corresponding to exchange area；
(4) by former subregion in the data syn-chronization after setting up snapshot into exchange area；
(5) former subregion is replaced with exchange area, defragmentation is completed.
After service inspection to some subregion needs to carry out disk arrangement, the index snapshot of the subregion is set up, will be above-mentioned fast
Effective mapping relations data write exchange area according in.
Snapshot portion arrangement has no effect on original subregion and persistently provides service, but can cause have new number when arranging snapshot
According to writing in original subregion, so a following step is that this partial data is synchronized in exchange area, and this partial data
It is very small for snapshot data, synchronizing process can be rapidly completed.
Exchange area is completing the arrangement of snapshot portion, and it is complete the new data of former subregion is also synchronously come after, then it is complete
Into the defragmentation of single subregion, former subregion is replaced.
Periodically remove
According to business demand, if certain group mapping relations no longer active (do not inquire about/change) within a period of time, such as
In 30 days, then need to remove these mapping relations from mapping library.In addition,
By periodically removing the size of data that also can effectively control to store in mapping library, it is to avoid data unrestrictedly expand,
And the influence brought after expansion to performance.
Because not being the deletion of business initiative, then requirement of real-time is not high.It is granularity in business using day based on this
The low ebb phase (morning), which concentrates, deletes expired mapping relations.
For how to determine whether one group of mapping relations is expired, an approximate algorithm is realized with Bloom filter,
Will not delete by mistake in the case of, performance and accuracy rate are also guaranteed.
First, major key identification code (including insertion, inquiry and change) the insertion cloth same day occurred by granularity of day is grand
Filter, and be stored in the form of a file on ordinary magnetic disc., can be grand by all cloth in the term of validity after regular removing starts
Filter is merged, and the filter after merging then covers all effective major key identification codes.Then Ergodic Maps are passed through
The mapping relations not appeared in filter are deleted in storehouse, until traversal is completed.
Performance test is reported
Performance test is carried out using SSD and Flash cards are (main), it is contemplated that major part ID is that do not have in inquiry request respectively
Mapping relations, so the ratio that the non-existent key and key of presence is inquired about in test is 2：1.
Mainly QPS, Latency (99%), CPU and disk read-write situation have been investigated in test.
| disk | disk | disk | Insert | Insert | query | query | |||
| r/s | rKB/s | util | load | CPU | lat99 | QPS | lat99 | QPS | Hardware |
| 11500 | 8500 | 90 | 6.9 | 198 | 0 | 2271 | 0 | 40000 | SSD |
| 26000 | 15420 | 84 | 7.4 | 260 | 0 | 3614 | 0 | 90000 | Flash cards |
| 29000 | 17000 | 87 | 7.8 | 295 | 0 | 2810 | 0 | 100000 | Flash cards |
| 31500 | 18700 | 93 | 7.3 | 314 | 0 | 2599 | 0 | 110000 | Flash cards |
| 35000 | 20450 | 97 | 7.6 | 364 | 0 | 837 | 0 | 120000 | Flash cards |
On end, CPU and disk utilization rate, and read speed and queryQPS it is proportional rise, query and insert
99 curves be always held at 0, can stablize handle all insertions and inquiry request.Insert with queryQPS rises can be by
Step declines, and may eventually reach the UPS upper performance score of hardware.The present invention can reach QPS in the case where carrying Flash cards
(100000-110000)。
Present invention performance under performance more preferably hardware also has preferable performance, such as carries Flash card ratios and uses
SSD performance boosts are more than 1 times.
In summary, compared with prior art, the present invention is wrapped a kind of method for setting up magnanimity ID mapping relations of the invention
Include and distribute unique identification code by different types for all ID, and mapping relations are stored in internal memory by identification code and indexed entirely
In；The identification code of major key in mapping relations selects corresponding subregion to be stored；When major key in the presence of, it is necessary to press
Identification code is selectively updated internal memory index and solid storage device entirely.Up to 1B is supported by above method unit of the present invention
(1,000,000,000) mapping relations are organized, expands storage capacity using multistage composite index and lifts query performance, pass through disk index partition
Mapping relations are recorded, mapping relations where realization is inquired about by any ID are indexed entirely by internal memory.The present invention has each party
Cookie and the mapping library per tripartite Cookie so that DSP can effectively be bidded according to Users'Data Analysis, realize profit
The maximization of benefit.Further DMP platforms (Data Management Platform.DMP stores flow, the various spies of audient
Reference ceases) just it is able to be integrated scattered first, third party's data, realize the unified identification of audient.
Embodiment described above is merely to illustrate the technological thought and feature of the present invention, in the art its object is to make
Technical staff can understand present disclosure and implement according to this, it is impossible to only limited with the present embodiment the present invention patent model
Enclose, i.e., all equal changes or modification made according to disclosed spirit still fall in the scope of the claims of the present invention.
Claims (2)
1. a kind of method for setting up magnanimity ID mapping relations, it is characterised in that comprise the following steps：
(1) the major key identification code pressed in mapping relations A selects corresponding block；
(2) judge whether current block there are the mapping relations, if so, step (3) is then performed, if it is not, then performing step (5)；
(3) the mapping relations B stored before the major key identification code is read；
(4) judge mapping relations B whether containment mapping relation A, if it is not, then perform step (5), if so, then perform step (7)；
(5) merge mapping relations A and B, and internal memory is updated to the mapping relations after one group of new mapping relations or merging
Full index；
(6) will mapping data write-in disk；And
(7) insertion mapping relations success,
Wherein, index is the index key of all mapping relations corresponding thereto to internal memory described in step (5) entirely.
2. the method as claimed in claim 1 for setting up magnanimity ID mapping relations, it is characterised in that：If pointed in index another
Outer major key identification code, then cover it, then this group of mapping relations are written in disk using current major key identification code.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410603292.8A CN104281717B (en) | 2014-10-31 | 2014-10-31 | A kind of method for setting up magnanimity ID mapping relations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410603292.8A CN104281717B (en) | 2014-10-31 | 2014-10-31 | A kind of method for setting up magnanimity ID mapping relations |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104281717A CN104281717A (en) | 2015-01-14 |
| CN104281717B true CN104281717B (en) | 2017-07-21 |
Family
ID=52256590
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410603292.8A Active CN104281717B (en) | 2014-10-31 | 2014-10-31 | A kind of method for setting up magnanimity ID mapping relations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104281717B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106656542B (en) * | 2015-11-04 | 2019-12-10 | 北京国双科技有限公司 | method and device for determining user identification |
| CN105743988B (en) * | 2016-03-30 | 2019-03-08 | 新浪网技术(中国)有限公司 | Network user's tracing implementing method, apparatus and system |
| CN108205575B (en) * | 2016-12-20 | 2021-10-08 | 北京国双科技有限公司 | Data processing method and device |
| CN108804695B (en) * | 2018-06-14 | 2022-05-13 | 广州谱道网络科技有限公司 | Promotion link generation and identification method and device |
| CN109828721B (en) * | 2019-01-23 | 2022-06-28 | 平安科技（深圳）有限公司 | Data deleting method and device, computer equipment and storage medium |
| CN110737747A (en) * | 2019-09-12 | 2020-01-31 | 苏宁云计算有限公司 | data operation method, device and system |
| CN116030398B (en) * | 2023-03-31 | 2023-06-23 | 山东爱特云翔计算机有限公司 | Task analysis method, system and storage medium based on multipath video signals |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1632765A (en) * | 2004-12-31 | 2005-06-29 | 大唐微电子技术有限公司 | A flash memory file system management method |
| CN101770515A (en) * | 2010-01-18 | 2010-07-07 | 杭州顺网科技股份有限公司 | Data block comparison based data updating method |
| EP2469423A1 (en) * | 2010-12-23 | 2012-06-27 | Sap Ag | Aggregation in parallel computation environments with shared memory |
| CN103166976A (en) * | 2013-04-03 | 2013-06-19 | 晶赞广告（上海）有限公司 | Synchronization cross-domain identify label mapping method of internet security |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110082855A1 (en) * | 2009-10-01 | 2011-04-07 | Al-Omari Awny K | Multi-dimensional access to data |
-
2014
- 2014-10-31 CN CN201410603292.8A patent/CN104281717B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1632765A (en) * | 2004-12-31 | 2005-06-29 | 大唐微电子技术有限公司 | A flash memory file system management method |
| CN101770515A (en) * | 2010-01-18 | 2010-07-07 | 杭州顺网科技股份有限公司 | Data block comparison based data updating method |
| EP2469423A1 (en) * | 2010-12-23 | 2012-06-27 | Sap Ag | Aggregation in parallel computation environments with shared memory |
| CN103166976A (en) * | 2013-04-03 | 2013-06-19 | 晶赞广告（上海）有限公司 | Synchronization cross-domain identify label mapping method of internet security |
Non-Patent Citations (2)
| Title |
|---|
| 对象存储设备端数据管理策略研究;刘景宁等;《计算机研究与发展》;20101231;第47卷(第10期);第1832-1839页 * |
| 支持大数据管理的NoSQL系统研究综述;申德荣等;《软件学报》;20131231;第24卷(第8期);第1786-1803页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104281717A (en) | 2015-01-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104281717B (en) | A kind of method for setting up magnanimity ID mapping relations | |
| CN105993013B (en) | A kind of data processing method apparatus and system | |
| US10303596B2 (en) | Read-write control method for memory, and corresponding memory and server | |
| CN101777016B (en) | Snapshot storage and data recovery method of continuous data protection system | |
| CN100399327C (en) | Managing file system versions | |
| CN101777017B (en) | Rapid recovery method of continuous data protection system | |
| WO2016086819A1 (en) | Method and apparatus for writing data into shingled magnetic record smr hard disk | |
| CN106649403B (en) | Index implementation method and system in file storage | |
| US20070005669A1 (en) | Method and system for automated disk i/o optimization of restored databases | |
| CN101582076A (en) | Data de-duplication method based on data base | |
| CN107798130A (en) | A kind of Snapshot Method of distributed storage | |
| CN102306168B (en) | Log operation method and device and file system | |
| CN107391544B (en) | Processing method, device and equipment of column type storage data and computer storage medium | |
| CN103473298B (en) | Data archiving method and device and storage system | |
| CN104536908B (en) | A kind of magnanimity small records efficient storage management method towards unit | |
| CN109407985B (en) | Data management method and related device | |
| CN110096509A (en) | Realize that historical data draws the system and method for storage of linked list modeling processing under big data environment | |
| CN106991190A (en) | A kind of database automatically creates subdata base system | |
| CN103473258A (en) | Cloud storage file system | |
| CN111708894B (en) | Knowledge graph creation method | |
| JP2012168781A (en) | Distributed data-store system, and record management method in distributed data-store system | |
| CN107506466A (en) | A kind of small documents storage method and system | |
| CN105183949A (en) | Railway main data cleaning method and system | |
| TW201021027A (en) | Disk layout method for object-based storage device | |
| CN102609364A (en) | Cache exchange method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |