CN105187502A - Method and system based on distributed elastic block storage - Google Patents
Method and system based on distributed elastic block storage Download PDFInfo
- Publication number
- CN105187502A CN105187502A CN201510484990.5A CN201510484990A CN105187502A CN 105187502 A CN105187502 A CN 105187502A CN 201510484990 A CN201510484990 A CN 201510484990A CN 105187502 A CN105187502 A CN 105187502A
- Authority
- CN
- China
- Prior art keywords
- mapping algorithm
- destination object
- file
- elastic block
- operated
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
Abstract
The invention relates to a method and a system based on distributed elastic block storage. The method comprises a step S1 of mapping a file to be operated into one or more target objects according to a first preset mapping algorithm, a step S2 of mapping the target objects into a placement group according to a second preset mapping algorithm, and a step S3 of mapping the placement group into an object storage cluster according to a third preset mapping algorithm; the system based on distributed elastic block storage comprises a first mapping module, a second mapping module and a third mapping module; the system based on distributed elastic block storage is a uniformed and distributed large-scale storage system designed for excellent performance, reliability and expandability; elastic block storage can be deployed in tens of thousands of servers, and then a uniformed block storage service is provided for customers.
Description
Technical field
The present invention relates to cloud computing and network field, particularly relate to a kind of method and system stored based on distributed elastic block.
Background technology
Along with the arrival in cloud epoch and the development of informationization technology, the static data such as such as picture, video, music and document constantly increases sharply in recent years, the demand of enterprise to memory capacity is increasing, and the task of top priority is the continuous dilatation realizing storing, and ensures the fast access to data.
The development course of telecommunications industry because of himself and the continuity of time, data environment more complicated, multiple storage mode is result in and the phenomenon of depositing in the different application of different times, larger telecommunications industry may have from DAS (DirectAttachedStorage simultaneously, i.e. direct-connected storage), NAS (NetworkAttachedStorage, namely network insertion stores) to the various storage organizations of SAN (StorageAreaNetwork, i.e. storage area network).
DAS refers to and is directly connected on a computer by memory device by scsi interface or optical-fibre channel, and operational format changes into file system and reoffers the local storage architecture used to application program.DAS is a kind of application memory technology comparatively early, be simple and easy to use, but disk utilization is very low, not easily dilatation.With under being usually used in single network environment and exchanges data amount is little, under the environment that performance requirement is not high.Jumbo data are stored and shares and be not suitable for.
SAN adopts optical-fibre channel (FibreChannel is called for short FC) technology, connects storage array and server host, set up the Local Area Network being exclusively used in data and storing by fibre channel media.SAN be applied in high to network speed requirement, to the reliability of data and security requirement is high, in the high applied environment of the performance requirement of data sharing, the big data quantity Mission critical applications of such as telecommunications, bank.It adopts block storage organization by providing high performance random I/O and data throughput at the data access of disk or FC level, it has the advantage of high bandwidth, low delay, but the price of SAN system is higher, and extensibility is poor, the continuous expansion of storage size can not be met.
NAS, by the network topology structure (such as Ethernet) of memory device by standard, is connected on a group computer.NAS is the storage means of component-level, it focus on help working group and department level mechanism to solve the demand of the memory capacity that increases sharply.NAS is that file-level stores, and it take file as host-host protocol, and realize networking by TCP/IP and store, extensibility is good, low price, user's manageability, but because the protocol overhead of NAS is high, bandwidth is low, delay is large, is unfavorable for applying in High-Performance Computing Cluster.
In sum, the different advantages of three kinds of technology make telecom operators respectively can get the chief when building different storage system, but it is not long-term plan that the complexity of traditional NAS and SAN storage array and high cost are all doomed these schemes for unstructured data, needs to be improved current storage organization.
Summary of the invention
Technical problem to be solved by this invention is to provide a kind of method and system stored based on distributed elastic block.
The technical scheme that the present invention solves the problems of the technologies described above is as follows: a kind of method stored based on distributed elastic block, comprises the following steps:
Step S1, is mapped as destination object by file to be operated according to default mapping algorithm one;
Step S2, is mapped in arrangement group by described destination object according to default mapping algorithm two;
Step S3, is mapped to described arrangement group in object storage cluster according to default mapping algorithm three.
On the basis of technique scheme, the present invention can also do following improvement.
Further, preset mapping algorithm one described in step S1 and comprise: the maximum capacity presetting described destination object, according to the maximum capacity of described destination object, cutting is carried out to described file to be operated and generate one or more described destination object.
Further, the described destination object generated after each cutting has unique recongnition of objects code, the generation method of described recongnition of objects code comprises: the document identifier code to be operated arranging file to be operated, be numbered by described file destination and make each described file destination have unique sequence number, described recongnition of objects code is obtained after described document identifier code to be operated by the sequence number cluster of described destination object.
Further, preset mapping algorithm two described in step S2 to comprise the following steps:
Step S21, uses static Hashing function to calculate the cryptographic Hash of described recongnition of objects code, described recongnition of objects code is become a pseudorandom values;
Step S22, by described pseudorandom values and mask step-by-step phase with, obtain the arrangement group sequence number that described destination object will map; Described mask is that the sum of described arrangement group subtracts one.
Further, preset mapping algorithm three described in step S3 and comprise: for settling the multiple described object storage cluster of group configuration described in each, namely the plurality of described object storage cluster is responsible for all destination objects in storage and maintenance arrangement group jointly.
The another kind of technical scheme that the present invention solves the problems of the technologies described above is as follows: a kind of system stored based on distributed elastic block, comprises the first mapping block, the second mapping block and the 3rd mapping block;
First mapping block is used for file to be operated to be mapped as destination object according to default mapping algorithm one;
Second mapping block is used for destination object to be mapped in arrangement group according to default mapping algorithm two;
3rd mapping block is used for arrangement group to be mapped in object storage cluster according to default mapping algorithm three.
On the basis of technique scheme, the present invention can also do following improvement.
Further, first mapping block comprises default mapping algorithm one unit, wherein said default mapping algorithm one comprises: the maximum capacity presetting described destination object, carries out cutting generate one or more described destination object according to the maximum capacity of described destination object to described file to be operated.
Further, the described destination object generated after each cutting has unique recongnition of objects code, the generation method of described recongnition of objects code comprises: the document identifier code to be operated arranging file to be operated, be numbered by described file destination and make each described file destination have unique sequence number, described recongnition of objects code is obtained after described document identifier code to be operated by the sequence number cluster of described destination object.
Further, the second mapping block comprises default mapping algorithm two unit, and described default mapping algorithm two unit comprises hash function computing unit and step-by-step phase and computing unit;
Described recongnition of objects code, for the cryptographic Hash using static Hashing function to calculate described recongnition of objects code, is become a pseudorandom values by hash function computing unit;
Step-by-step phase and computing unit be used for by described pseudorandom values and mask step-by-step phase with, obtain the arrangement group sequence number that described destination object will map; Described mask is that the sum of described arrangement group subtracts one.
Further, 3rd mapping block comprises default mapping algorithm three unit, wherein said default mapping algorithm three comprises: for settling the multiple described object storage cluster of group configuration described in each, and namely the plurality of described object storage cluster is responsible for all destination objects in storage and maintenance arrangement group jointly.
The invention has the beneficial effects as follows:
1, Single-Server bottleneck and fault point is solved: use distributed computing technology, easily can meet I/O intensive applications scene; Elastomer block stores and uses multiple copies technology, data is copied on multiple servers in real time, even if we cannot avoid hard disk failure, still can ensure persistent data.
2, adopt common server to substitute disk array: traditional disk array has costliness, inflexible feature, elastomer block stores does not exist this problem.Memory capacity can easily realize expansion by adding memory node.
3, more powerful High Availabitity, error detection and repair function is provided: disk array is once bad two pieces of hard disc datas will be lost, and the Data distribution8 formula of elastomer block is on the different hard disks of different servers, and reliability is stronger.
4, the performance of described elastomer block storage system linearly can promote with the expansion of memory capacity.
Accompanying drawing explanation
Fig. 1 is the method overview flow chart stored based on distributed elastic block of the present invention;
Fig. 2 is the method particular flow sheet stored based on distributed elastic block of the present invention;
Fig. 3 is the method data write flow chart stored based on distributed elastic block of the present invention;
Fig. 4 is the system construction drawing stored based on distributed elastic block of the present invention;
Fig. 5 is the structural framing figure stored based on distributed elastic block of the present invention.
Embodiment
Be described principle of the present invention and feature below in conjunction with accompanying drawing, example, only for explaining the present invention, is not intended to limit scope of the present invention.
Fig. 1 is the method overview flow chart stored based on distributed elastic block of the present invention.
As shown in Figure 1, a kind of method stored based on distributed elastic block, comprises the following steps:
Step S1, is mapped as one or more destination object by file to be operated according to default mapping algorithm one;
Step S2, is mapped in arrangement group by destination object according to default mapping algorithm two;
Step S3, is mapped to arrangement group in object storage cluster according to default mapping algorithm three.
Fig. 2 is the method particular flow sheet stored based on distributed elastic block of the present invention.
As shown in Figure 2, step S1 is from File and the file-object to be operated i.e. mapping of destination object, and the current object mapped is the file that user will be operated, and is mapped as accessible object.
Preset mapping algorithm one to comprise: the maximum capacity presetting object, then carry out cutting according to the maximum capacity of object to file, this step is equivalent to the striping process in RAID.The beneficial effect of step S11 comprises: the file that first size can be allowed not limit becomes multiple object that heap(ed) capacity is consistent, be convenient to efficient management; Next is that the serial process implemented single file becomes the parallelization process implemented multiple object.
The object produced after each cutting will obtain unique recongnition of objects code (oid, i.e. objectid).The producing method of Oid is Linear Mapping.Be specially: arranging ino is document identifier code to be operated, is the unique identifier of file; Be numbered by the object produced by file cutting, suppose that ono is the sequence number of certain object produced by file cutting, so ono simply joins together and obtains after ino by the oid of this object.
Step S2 is the mapping from Object-arrangement group and PG (PlacementGroup); After file is mapped as one or more object, each object is mapped to independently in a PG and goes.
Preset mapping algorithm two to comprise the following steps:
Step S21, uses static Hashing function to calculate the cryptographic Hash of oid, oid is become the pseudorandom values of an approaches uniformity distribution.
Step S22, by pseudorandom values and mask step-by-step phase with, be rehoused group sequence number and pgid.
Above-mentioned steps S21 and step S22 calculating formula table show and are: hash (oid) & mask-pgid; Wherein PG add up to m (m is the integer power of 2), then the value of mask is m-1.Therefore, the whole result of cryptographic Hash calculating and step-by-step and operation is in fact approaches uniformity ground Stochastic choice one from all m PG.Based on this mechanism, when there being a large amount of object and a large amount of PG, can ensure that the approaches uniformity between object and PG maps.And because object is by file cutting, the size of most of object is identical, thus, this mapping can ensure the total amount of data approaches uniformity distribution of the object stored in each PG.
Step S3 is from PG-OSD (objectstoragedevice) the i.e. mapping of object storage cluster.
Preset mapping algorithm three to comprise: the physical memory location OSD PG of the logical organization unit as object being mapped to data, a PG can be mapped on n OSD, and each OSD can carry a large amount of PG, be namely " multi-to-multi " mapping relations between PG and OSD.Then obtain one group of altogether n OSD, namely the numerical value of all object, the n of this n OSD jointly in responsible storage and maintenance PG can configure, value 3 under normal conditions for the demand of reliability according in practical application.Specific to each OSD, be then responsible for performing the operation such as the storage of object in local file system, access, metadata maintenance being mapped to this locality by the OSDdeamon that it runs.
By the cubic map of above-mentioned steps S1, S2 and S3, complete from file to object, the whole mapping process of PG and OSD.
Below with a specific embodiment, data write flow process is described.
Fig. 3 is the method data write flow chart stored based on distributed elastic block of the present invention.
As shown in Figure 3, when user and client write a file, be first an object by file cutting, then find out one group of three OSD storing this object.Encoding to these three OSD makes these three OSD have different sequence numbers, and arranging the most forward OSD of sequence number is PrimaryOSD, and other two are followed successively by SecondaryOSD and TertiaryOSD.After finding out three OSD, client directly communicates with PrimaryOSD, initiates write operation (as Write1 in Fig. 3); After PrimaryOSD receives request, initiate write operation (as Write2 and Write3 in Fig. 3) respectively to SecondaryOSD and TertiaryOSD; After SecondaryOSD and TertiaryOSD completes write operation separately, send confirmation (as Ack4 and Ack5 in Fig. 3) respectively to PrimaryOSD; After PrimaryOSD be sure of that the write of other two OSD completes, then oneself also complete data write, and confirm that object write operation completes (as Ack6 in Fig. 3) to client.
Adopt above-mentioned write flow process, the reliability in ablation process can be ensured, avoid as far as possible causing loss of data.
Fig. 4 is the system construction drawing stored based on distributed elastic block of the present invention.
As shown in Figure 4, a kind of system stored based on distributed elastic block, comprises the first mapping block, the second mapping block and the 3rd mapping block;
First mapping block is used for file to be operated to be mapped as one or more destination object according to default mapping algorithm one;
Second mapping block is used for destination object to be mapped in arrangement group according to default mapping algorithm two;
3rd mapping block is used for arrangement group to be mapped in object storage cluster according to default mapping algorithm three.
The system stored based on distributed elastic block of the present invention also comprises cluster monitor, for monitoring and manage the mapping of the first mapping block, the second mapping block and the 3rd mapping block.
First mapping block comprises default mapping algorithm one unit, and wherein said default mapping algorithm one comprises: the maximum capacity presetting object, the maximum capacity according to object carries out cutting to file.The object produced after each cutting has unique recongnition of objects code, the generation method of oid comprise for: arranging ino is document identifier code to be operated, be numbered by object and make each object have unique sequence number, oid is obtained after ino by the sequence number cluster of object.
Second mapping block comprises default mapping algorithm two unit, and described default mapping algorithm two unit comprises hash function computing unit and step-by-step phase and computing unit; Oid, for the cryptographic Hash using static Hashing function to calculate oid, is become a pseudorandom values by hash function computing unit; Step-by-step phase and computing unit be used for by pseudorandom values and mask step-by-step phase with, be rehoused group sequence number and pgid.
3rd mapping block comprises default mapping algorithm three unit, and wherein said default mapping algorithm three comprises: bring pgid into according to algorithm, and obtain one group of n OSD altogether, namely this n OSD is responsible for all object in storage and maintenance PG jointly.
The system stored based on distributed elastic block of the present invention is a kind of is outstanding performance, reliability and extensibility and unification, the distributed large-scale storage system that designs.Elastomer block storage can be deployed in on front station server up to ten thousand, then provides block stores service to client is unified.
The method and system stored based on distributed elastic block of the present invention, has following characteristic:
High reliability: being first the data for storing in systems in which, can ensureing that data can not be lost; Secondly, the reliability in data writing process is high, namely user by data writing process, loss of data can not be caused because of the appearance of fortuitous event.
Increasingly automated: the automatic renewal (replication) specifically including data, autobalance (re-balancing), automatic error detection (failuredetection) and automatic errors repair (failurerecovery).In general, these automation characteristics ensure that the height of system is reliable on the one hand, have also ensured that its O&M difficulty still can remain on a relatively low level after system scale expands on the one hand.
Enhanced scalability: " easily extensible " here had both included the easily extensible of system scale and memory capacity, also include the linear expansion of the aggregated data access bandwidth along with the increase of system node number, further comprises and provide several functions based on the Basic API that feature richness is powerful, support the functional easily extensible of multiple application.
Fig. 5 is the structural framing figure stored based on distributed elastic block of the present invention.
As shown in Figure 5, described elastomer block storage system is applicable to dynamic and static state data, and for storing documents, virtual machine image, application and database data, elastomer block storage system may be used in cloud infrastructure system.Embodiment is as follows:
1, geographically isolated by machine, machine is placed on 3 racks respectively; Each rack has 3 storage servers, and each storage server has three OSD memory nodes, adopts identical configuration;
2, the number of copies of configuring redundancy is 3, on the memory node of the different storage servers of 3 different frames, namely has the data copy existence that intact, to guarantee data security;
3, generate cluster topology information, store frame, storage server, memory node and OSD, the map informations such as configuration group and cluster, and map information is distributed on each storage server, start elastomer block storage cluster;
4, be deployed in local use on the application server, method is identical with local hard drive using method.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1., based on the method that distributed elastic block stores, it is characterized in that, comprise the following steps:
Step S1, is mapped as destination object by file to be operated according to default mapping algorithm one;
Step S2, is mapped in arrangement group by described destination object according to default mapping algorithm two;
Step S3, is mapped to described arrangement group in object storage cluster according to default mapping algorithm three.
2. the method stored based on distributed elastic block according to claim 1, it is characterized in that, preset mapping algorithm one described in step S1 to comprise: the maximum capacity presetting described destination object, according to the maximum capacity of described destination object, cutting is carried out to described file to be operated and generate one or more described destination object.
3. the method stored based on distributed elastic block according to claim 2, it is characterized in that, the described destination object generated after each cutting has unique recongnition of objects code, the generation method of described recongnition of objects code comprises: the document identifier code to be operated arranging file to be operated, be numbered by described file destination and make each described file destination have unique sequence number, described recongnition of objects code is obtained after described document identifier code to be operated by the sequence number cluster of described destination object.
4. the method stored based on distributed elastic block according to claim 3, is characterized in that, presets mapping algorithm two and comprise the following steps described in step S2:
Step S21, uses static Hashing function to calculate the cryptographic Hash of described recongnition of objects code, described recongnition of objects code is become a pseudorandom values;
Step S22, by described pseudorandom values and mask step-by-step phase with, obtain the arrangement group sequence number that described destination object will map; Described mask is that the sum of described arrangement group subtracts one.
5. the method stored based on distributed elastic block according to claim 4, it is characterized in that, preset mapping algorithm three described in step S3 to comprise: for settling the multiple described object storage cluster of group configuration described in each, namely the plurality of described object storage cluster is responsible for all destination objects in storage and maintenance arrangement group jointly.
6., based on the system that distributed elastic block stores, it is characterized in that, comprise the first mapping block, the second mapping block and the 3rd mapping block;
Described first mapping block is used for file to be operated to be mapped as destination object according to default mapping algorithm one;
Described second mapping block is used for destination object to be mapped in arrangement group according to default mapping algorithm two;
Described 3rd mapping block is used for arrangement group to be mapped in object storage cluster according to default mapping algorithm three.
7. the system stored based on distributed elastic block according to claim 6, it is characterized in that, described first mapping block comprises default mapping algorithm one unit, wherein said default mapping algorithm one comprises: the maximum capacity presetting described destination object, carries out cutting generate one or more described destination object according to the maximum capacity of described destination object to described file to be operated.
8. the system stored based on distributed elastic block according to claim 7, it is characterized in that, the described destination object generated after each cutting has unique recongnition of objects code, the generation method of described recongnition of objects code comprises: the document identifier code to be operated arranging file to be operated, be numbered by described file destination and make each described file destination have unique sequence number, described recongnition of objects code is obtained after described document identifier code to be operated by the sequence number cluster of described destination object.
9. the system stored based on distributed elastic block according to claim 8, it is characterized in that, described second mapping block comprises default mapping algorithm two unit, and described default mapping algorithm two unit comprises hash function computing unit and step-by-step phase and computing unit;
Described recongnition of objects code, for the cryptographic Hash using static Hashing function to calculate described recongnition of objects code, is become a pseudorandom values by described hash function computing unit;
Described step-by-step phase and computing unit be used for by described pseudorandom values and mask step-by-step phase with, obtain the arrangement group sequence number that described destination object will map; Described mask is that the sum of described arrangement group subtracts one.
10. the system stored based on distributed elastic block according to claim 9, it is characterized in that, described 3rd mapping block comprises default mapping algorithm three unit, wherein said default mapping algorithm three comprises: for settling the multiple described object storage cluster of group configuration described in each, and namely the plurality of described object storage cluster is responsible for all destination objects in storage and maintenance arrangement group jointly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510484990.5A CN105187502A (en) | 2015-08-07 | 2015-08-07 | Method and system based on distributed elastic block storage |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510484990.5A CN105187502A (en) | 2015-08-07 | 2015-08-07 | Method and system based on distributed elastic block storage |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105187502A true CN105187502A (en) | 2015-12-23 |
Family
ID=54909342
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510484990.5A Pending CN105187502A (en) | 2015-08-07 | 2015-08-07 | Method and system based on distributed elastic block storage |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105187502A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106503237A (en) * | 2016-11-04 | 2017-03-15 | 郑州云海信息技术有限公司 | A kind of PG member selections method, device, storage pool and distributed file system |
| CN106649702A (en) * | 2016-12-20 | 2017-05-10 | 上海斐讯数据通信技术有限公司 | File storage method and apparatus of cloud storage system, and cloud storage system |
| WO2017107948A1 (en) * | 2015-12-23 | 2017-06-29 | 中兴通讯股份有限公司 | File write-aggregation and read-aggregation method and system, and client |
| CN108052655A (en) * | 2017-12-28 | 2018-05-18 | 新华三技术有限公司 | Data write and read method |
| CN109542622A (en) * | 2018-11-21 | 2019-03-29 | 新华三技术有限公司 | A kind of data processing method and device |
| CN111026337A (en) * | 2019-12-30 | 2020-04-17 | 中科星图股份有限公司 | Distributed storage method based on machine learning and ceph thought |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103067488A (en) * | 2012-12-25 | 2013-04-24 | 中国科学院深圳先进技术研究院 | Implement method of unified storage |
| CN104243564A (en) * | 2014-09-03 | 2014-12-24 | 河海大学 | Distributed data management method for AWS platform |
-
2015
- 2015-08-07 CN CN201510484990.5A patent/CN105187502A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103067488A (en) * | 2012-12-25 | 2013-04-24 | 中国科学院深圳先进技术研究院 | Implement method of unified storage |
| CN104243564A (en) * | 2014-09-03 | 2014-12-24 | 河海大学 | Distributed data management method for AWS platform |
Non-Patent Citations (1)
| Title |
|---|
| 叶为民: "分布式文件系统自适应负载调度研究", 《中国优秀硕士学位论文数据库（信息科技辑）》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017107948A1 (en) * | 2015-12-23 | 2017-06-29 | 中兴通讯股份有限公司 | File write-aggregation and read-aggregation method and system, and client |
| CN106503237A (en) * | 2016-11-04 | 2017-03-15 | 郑州云海信息技术有限公司 | A kind of PG member selections method, device, storage pool and distributed file system |
| CN106503237B (en) * | 2016-11-04 | 2019-09-24 | 郑州云海信息技术有限公司 | A kind of PG member selection method, apparatus, storage pool and distributed file system |
| CN106649702A (en) * | 2016-12-20 | 2017-05-10 | 上海斐讯数据通信技术有限公司 | File storage method and apparatus of cloud storage system, and cloud storage system |
| CN108052655A (en) * | 2017-12-28 | 2018-05-18 | 新华三技术有限公司 | Data write and read method |
| CN109542622A (en) * | 2018-11-21 | 2019-03-29 | 新华三技术有限公司 | A kind of data processing method and device |
| CN111026337A (en) * | 2019-12-30 | 2020-04-17 | 中科星图股份有限公司 | Distributed storage method based on machine learning and ceph thought |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11249857B2 (en) | Methods for managing clusters of a storage system using a cloud resident orchestrator and devices thereof | |
| US10228871B2 (en) | Enabling data integrity checking and faster application recovery in synchronous replicated datasets | |
| CN105187502A (en) | Method and system based on distributed elastic block storage | |
| US10769024B2 (en) | Incremental transfer with unused data block reclamation | |
| JP5026283B2 (en) | Collaborative shared storage architecture | |
| CN103503414B (en) | A kind of group system calculating storage and merge | |
| US10852984B2 (en) | Mirror vote synchronization | |
| US10353870B2 (en) | Tracking structure for data replication synchronization | |
| CN109313538B (en) | Inline deduplication | |
| CN109690494B (en) | Hierarchical fault tolerance in system storage | |
| US20200201722A1 (en) | Cross-platform replication | |
| US11836115B2 (en) | Gransets for managing consistency groups of dispersed storage items | |
| CN103929500A (en) | Method for data fragmentation of distributed storage system | |
| US20160050146A1 (en) | Accelerated storage appliance using a network switch | |
| CN104050102B (en) | Object storage method and device in a kind of telecommunication system | |
| US10031682B1 (en) | Methods for improved data store migrations and devices thereof | |
| US20180275907A1 (en) | Composite aggregate architecture | |
| US11768624B2 (en) | Resilient implementation of client file operations and replication | |
| KR101912728B1 (en) | Data processing method and apparatus | |
| US20150381727A1 (en) | Storage functionality rule implementation | |
| CN110663034B (en) | Method for improved data replication in cloud environment and apparatus therefor | |
| KR102084031B1 (en) | Method for managing integrated local storage and apparatus therefor | |
| WO2018235132A1 (en) | Distributed storage system | |
| US20190332293A1 (en) | Methods for managing group objects with different service level objectives for an application and devices thereof | |
| US11327844B1 (en) | Automated cloud recovery to provide a full usable application image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151223 |
|
| RJ01 | Rejection of invention patent application after publication |