CN103124299A - Distributed block-level storage system in heterogeneous environment - Google Patents
Distributed block-level storage system in heterogeneous environment Download PDFInfo
- Publication number
- CN103124299A CN103124299A CN2013100917711A CN201310091771A CN103124299A CN 103124299 A CN103124299 A CN 103124299A CN 2013100917711 A CN2013100917711 A CN 2013100917711A CN 201310091771 A CN201310091771 A CN 201310091771A CN 103124299 A CN103124299 A CN 103124299A
- Authority
- CN
- China
- Prior art keywords
- node server
- piece
- block
- storage system
- distributed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a distributed block-level storage system in a heterogeneous environment. The distributed block-level storage system in the heterogeneous environment is characterized in that a virtual block device is abstracted from multiple low-priced server devices which serve as block node servers by utilizing a storage virtualization technology; and an agent node server with a high availability mode is set up to manage the block node servers and forward I/O (Input/Output) requests sent out by a user. The block node servers improve flexibility of scale change of the node servers by using a distributed virtual node distribution policy based Consistent Hash data distribution algorithm and improve the whole performance of the system by introducing a node capacity perception based load balancing method. According to the requirement of massive scale storage environment, the distributed block-level storage system in the heterogeneous environment is used for improving utilization rate of a storage space of a host in a data center; and block-level storage spaces after the virtualization can be distributed to application servers of various users as required.
Description
Technical field
The present invention relates to a kind of storage system, relate in particular to the distributed block rank storage system under a kind of isomerous environment.
Background technology
Traditional storage system adopts the method for centralised storage, and the method is all data of storage on single storage server, the read-write requests of being responsible for replying the user.The method of centralised storage requires storage server to possess jumbo characteristic, can store the data of a large number of users, and require specific file system to move on it, directly read to facilitate the user, yet readwrite performance reduces when userbase enlarges greatly.
In recent years, most storage system is all that storage system is located in peripheral equipment with server and user-center.But along with the quickening of social informatization process, amount of information sharply increases, and webpage quantity is huge, and the data capacity of the storage of each enterprise is the geometric progression amount and increases.Traditional centralised storage method can normally be used in the little information system of data volume.Yet when data volume increased on a large scale, its readwrite performance became the bottleneck of information system, and its reliability and fail safe can not be satisfied the needs of application.And the most storage system all is based on file system, and the read-write of data must be based on file system, and this makes the file system read-write of data that become bottleneck effect.
Distributed memory system disperses data to be stored on the physical storage device of many platform independent exactly.Distributed memory system adopts extendible system configuration, utilizes many piece node servers to share the storage load, utilizes agent node server selection storage information, and it has not only improved reliability, availability and the access efficiency of system, also is easy to expansion.
Along with the development of computer hardware and the sharply increase of amount of information, tradition adopts the memory technology of centralised storage method can not effectively utilize the advantage of a large amount of cheap hardware, and its utilization to the performance of machine own is also insufficient.Along with the continuous popularization of distributed computing technology and Intel Virtualization Technology, also make the distributed virtualization technology become the effective ways that improve performance of storage system, and can provide the storage of piece rank to the user.Comparatively speaking, traditional centralised storage and file-level storage list reveal on performance greatly limitation.
Summary of the invention
For the problem that exists in above-mentioned traditional storage system, need to propose a kind ofly can take full advantage of computing power and low-cost server hardware, and utilize the storage system of distributed computing technology.This storage system should have that memory capacity is large, redundancy good, the function of high availability, makes under a large amount of requests of multi-user's situation and still has readwrite performance preferably, preferably disaster tolerance.By utilizing the advantage of low-cost server cluster and Intel Virtualization Technology, promote memory capacity, take full advantage of computer hardware and Internet resources, improve load equilibrium, solve the problem of heritage storage system I/O inefficiency under large-scale data.
The present invention pays close attention to traditional storage system weak point, utilizes storage virtualization technology that the storage system of the distributed block rank under a kind of isomerous environment is provided.Described storage system comprises: the agent node server, directly be connected with the user, and the mapping between providing from the virtual memory facilities to the physical storage device to the user; The piece node server communicates with described acting server, and the amount of physical memory take piece as unit is provided.
Concrete ins and outs of the present invention comprise:
(1) the piece node server is erected on the physical store environment of isomery.
The hardware infrastructure framework is on large-scale low-cost server cluster.Powerful from traditional performance but expensive large-scale computer is different, architecture of the present invention has been used cheap server cluster in a large number, the server of x86 framework particularly, and the internet between server generally also uses general gigabit Ethernet.
Utilize storage virtualization technology, shield concrete hardware resource details, form unified resource pool, fully transparent to the user, give concrete the application according to the virtualized storage resources of demand assignment.The basic conception of Storage Virtualization is that the physical store entity of reality and the logical expressions of storage are separated, and a plurality of memory devices are obtained unified management in a storage pool.Main frame on IP network is only mutual with the logical volume of distributing to them, and is indifferent to its deposit data on that physical store entity.
By the redundancy between a plurality of low-cost servers, make the bottom data storage obtain high availability.Owing to having used cheap server cluster.The inefficacy of server will be inevitable, and have the problem that server lost efficacy simultaneously.For this reason, the present invention uses the server of redundancy to obtain high availability at the Fault-Tolerant Problems that needs on Software for Design to consider between server.
(2) possesses the agent node server of high enabled mode.
Be responsible for forwarding user's read-write requests due to the agent node server, it must possess high availability.The agent node server adopts the Master/Standby pattern to prevent the Single Point of Faliure that lost efficacy and bring due to it, and Master agent node server provides normal service, and Standby agent node server is as secondary node.In case Master agent node server breaks down, the Standby node server can initiatively be taken over the storage system interactive service, guarantees the system business continuity
Its high enabled mode comprises three major functions:
1 redundancy feature, i.e. the redundancy scheme of single failpoint in elimination system, cluster information synchronous.
2 function of failure detections namely can in time detect by mechanism such as transmission of messages the function of component failure, and the present invention adopts heartbeat (Heartbeat) pattern directly to carry out failure detection, directly sends message to the other side between the agent node server.Message Adoption Network connection mode sends, monitoring server state and service running status.System possesses the agent node server and surveys and restore funcitons to the multipoint fault between (disk heartbeat) many paths between the piece node server.
3 failover functions are namely served the function of switching with redundant resource after thrashing being detected.
(3) the piece node server of distributed management
The piece node server provides the service of iSCSI Target, WWN and unique sign and the agent node server of LUN of rolling up ID, Target and Initiator passage by this locality when each piece node server starts are set up mapping relations, be conducive to Information Security, prevent unauthorized access, if can't connect with the agent node server, the service meeting autoshutdown of piece node server so.
The piece node server can be stated the block copy that it has by sending a block report to the agent node server.The information of block report comprises the id of this block, the block copy length that timestamp (time stamp) and it are held.After completing, the registration of piece node server will send immediately primary block report.Can just carry out every 1 hour afterwards block reports and send, thereby provide latest position information about all the block copies in this cluster for the agent node server.
Under normal circumstances, it is available with the block copy that confirms it oneself is moving and it is held that the piece node server can send heartbeat message to the agent node server, and the heart beat cycle of acquiescence was 3 seconds.If the agent node server can not receive the heartbeat message from certain piece node server in ten minutes, it can think that this piece node server can not provide service, and the block copy that it is held has just become down state.The agent node server will creating out on other piece node servers these block copy self-adaptings.
From also carrying more current information about total memory capacity, memory space use amount and the current volume of transmitted data aspect of processing in the heartbeat of piece node server.These statistical informations can be used in the allocation of space and load balance decision of agent node server.The agent node server can not contacted directly the piece node server, but by the response message of heartbeat is come to send instruction to the piece node server:
1. delete local block copy
2. re-register or close the piece node server
3. send an instant block report
The present invention adopts the main purpose of consistency Hash in the management block node server be when changing the Node quantity of cluster, can change as few as possible the mapping relations that have Key and piece node server.The thinking of this algorithm is divided into following three steps.At first calculate the cryptographic Hash of each piece node server, and it is assigned to one 0 ~ 2
32Annular regions between on.Next uses same procedure to calculate the cryptographic Hash of the Block that the agent node server forwards, also it is assigned on this annulus.Subsequently from data-mapping to the position begin to search clockwise, data are saved on first piece node server that finds.If surpass 2
32Still can not find the piece node server, will be saved on first piece node server.Suppose to have 5 piece node servers in this annular Hash space, if increase a piece node server 6, draw piece node server 6 according to hash algorithm and be mapped between piece node server 3 and piece node server 4, so affected will be only to traverse counterclockwise object between piece node server 3 along piece node server 6.
(4) based on the load balancing of the distribution method of piece node server capacity perception
The realization of this load balancing is based on following hypothesis: the cryptographic Hash of storage object is evenly distributed, and the storage overhead of each memory address section is evenly distributed.The basic thought of optimizing is that the storage resources of each piece node server is estimated to quantize, and according to the resource situation of whole system, the dummy node number that the piece node server distributes is adjusted, thereby is further optimized load balancing between the piece node server.Distribute regulatory factor w for each piece node server, the resource of this piece node server of this Parametric Representation is distributed weights, can represent memory capacity or bandwidth etc., and what adopt in Prototype Design is quantization weight to capacity resource.Then, add up each piece node server distribute the big or small s of address space.Based on hypotheses, the s value can represent storage overhead; If P=s/w is as the resource utilization quantized value.System obtains the average of P by statistical computation, and sets threshold value P
HighAnd P
lowWhen node surpasses threshold range, adopt the method that increases or reduce the dummy block will node server, the load of balance whole system.
When initiating read-write requests by the user, request is distributed to according to load balancing and completes actual read-write requests on the piece node server by the agent node server.Distributed memory system according to the present invention is based on physical storage device (for example, disk), namely the piece level other.So no matter be what operating system or file system, can include easily management in, and data be by this one deck of file system, thereby accelerated the speed of transfer of data.
The beneficial effect that the present invention has is:
1, the present invention has used the server of a large amount of cheapnesss, utilizes storage virtualization technology, shields concrete hardware resource details, forms unified memory resource pool, has that capacity is large, characteristics of redundancy and high availability.
2, the present invention utilizes the high enabled mode of agent node server, guarantees the stability of a system, guarantees the continuity of system read-write
3, the present invention utilizes consistency Hash Distribution Algorithm to reduce the increase of piece node server and deleted the data distributions is brought impact, be conducive to simultaneously the better load of group system and dispersiveness, and the difference of considering each piece node server storage capacity under isomerous environment.
4, utilization of the present invention is based on the distribution method of piece node server capacity perception, for the reasonable utilization of system storage resource provides mechanism more flexibly, guaranteed the load balancing of system storage resource.
Description of drawings
Fig. 1 is the schematic diagram of the system architecture configuration of storage system of the present invention.
Fig. 2 is the flow chart that storage system of the present invention is read and write.
Embodiment
The present invention is described in further detail below in conjunction with accompanying drawing and implementation method.
With reference to Fig. 1, hardware configuration according to the piece level distributed memory system of the embodiment of the present invention is described.Fig. 1 is the schematic diagram of having described according to the system architecture configuration of the piece level distributed memory system of the embodiment of the present invention.
Describe as Fig. 1, comprise 2 agent node servers 1000 according to the piece level distributed memory system of the embodiment of the present invention, and the different physical storage device of many performances that provided by different vendor is as piece node server 2000.
Agent node server, piece node server are described as Fig. 1.Yet, it will be appreciated by persons skilled in the art that as required, the number of these building blocks can suitably be set, and the present invention is not limited thereto.
Here, offer user's mapping between (for example, disk) from the virtual memory facilities to the physical storage device according to the agent node server of the embodiment of the present invention, and can manage various disk arrays (RAID) or physical storage device.
In addition, be to provide amount of physical memory take piece as unit according to the effect of the block server of the embodiment of the present invention, for example 128MB/ is every.Certainly, it will be appreciated by persons skilled in the art that according to actual needs, block size can at random be arranged.
With reference to Fig. 2 execution in step, flow chart of the invention process is described:
(1) system's deployment
At first, such as step 11 description, the agent node server with high enabled mode is set.The agent node server is 2 station servers, comprises storage pool and manages this module, and it is managed concentratedly storage resources, the storage resources that is dispersed in each server or isomery can be included in as required and be carried out unified centralized management in one or more ponds.Also have the virtual volume management, this module realizes the resource in storage pool is undertaken by the stored configuration demand logical separation of effective and reasonable storage resources by storage pool being divided into a plurality of virtual volumes (logical volume).With the mapping of LUN Mapping(LUN) module, this module realizes the use of virtual volume (logical volume) is distributed, and storage resources is provided for various application servers or main frame by FC and ISCSI dual mode.And the Master/Standby pattern is set has reached high availability.Then change step 12 over to and dispose the distributed block node server, the piece node server is many cheap heterogeneous servers, distribute and dispose based on the capacity perception algorithm of node and complete according to the consistency Hash, and on each piece node server, the disk management module being installed, can carry out basic management to the disk in system, the disk index is provided, the disk total capacity, disk active volume, Disk State, whether disk the information such as initialization, can also carry out initialization to the no initializtion disk.Enter step 13 and judge whether that new piece node server adds, enter step 14 if having, adopt the consistency hash algorithm to calculate the cryptographic Hash of each piece node server, and it is assigned to one 0 ~ 2
32Annular regions between on.Next uses same procedure to calculate the cryptographic Hash of the piece that the agent node server forwards, and also it is assigned on this annulus.Subsequently from data-mapping to the position begin to search clockwise, data are saved on first node that finds.If surpass 2
32Still can not find node, will be saved on first node.If add without new piece node server, enter step 15, load balancing is set.Be that the resource of each piece node server is estimated to quantize, according to the resource situation of whole system, the dummy node number that the piece node server distributes adjusted, thus the load balancing between optimization piece node server.Distribute regulatory factor w for each piece node server, the resource of this piece node server of this Parametric Representation is distributed weights, can represent the capacity of node or bandwidth etc., and what adopt in Prototype Design is quantization weight to capacity resource.Then, add up each piece node server distribute the big or small s of address space.Based on hypotheses, the s value can represent this node institute memory allocated expense; If P=s/w is as piece node server resource utilization quantized value.System obtains the average of P by statistical computation, and sets threshold value P
HighAnd P
lowWhen node surpasses threshold range, adopt the method that increases or reduce dummy node, balanced whole system load.Describe as step 16, all node deployments are completed, and change (2) over to.
(2) response user request process.
After the request that receives the user, by the agent node server, enter step 17, judged whether that the user sends read-write requests, if without flow process end.Otherwise enter step 18, request will arrive the agent node server.Then change step 19 over to, it is read request or write request that the agent node server is judged as the request that the user sends.If request enters step 20 for read request, the agent node server reads data and returns to the user from different piece node servers.If request enters step 21 for write request, the agent node server is pressed block size, stores into randomly in the different masses node server.
Therefore, reading all data with traditional network store system employing from the storage server of concentrating compares, can read and write randomly, dispersedly corresponding data according to the piece level distributed memory system of the embodiment of the present invention from the dissimilar physical storage device of being made by different vendor, thereby realize the management of the memory space take piece as unit.
Claims (5)
1. the distributed block rank storage system under an isomerous environment, is characterized in that: under the isomerism storage resources environment, set up agent node server and distributed node server, piece rank memory space externally is provided; Both sides relation is: the agent node server forwards user's read-write requests to the piece node server, and the piece node server provides the load balancing of resource to the agent node server.
2. require distributed block rank storage system under described a kind of isomerous environment as right 1, it is characterized in that: described isomerous environment is that the amount of physical memory that the piece node server provides derives from dissimilar memory device, and quantity is a plurality of.
3. require distributed block rank storage system under described a kind of isomerous environment as right 1, it is characterized in that: the agent node Servers installed becomes high enabled mode, it accepts the read-write operation that the user sends, computation requests is to the physical address map information of piece node server, and these read-write operations to logical volume are converted into read-write operation to the actual physics dish.
4. require distributed block rank storage system under described a kind of isomerous environment as right 1, it is characterized in that: the piece node server can manage its disk; But and adopt consistency Hash Distribution Algorithm to build the distributed block node server of additions and deletions.
5. require distributed block rank storage system under described a kind of isomerous environment as right 1, it is characterized in that: the storage resources to each piece node server quantizes, piece node server resource situation according to whole system realizes the load balancing between the piece node server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013100917711A CN103124299A (en) | 2013-03-21 | 2013-03-21 | Distributed block-level storage system in heterogeneous environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013100917711A CN103124299A (en) | 2013-03-21 | 2013-03-21 | Distributed block-level storage system in heterogeneous environment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103124299A true CN103124299A (en) | 2013-05-29 |
Family
ID=48455149
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013100917711A Pending CN103124299A (en) | 2013-03-21 | 2013-03-21 | Distributed block-level storage system in heterogeneous environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103124299A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103929500A (en) * | 2014-05-06 | 2014-07-16 | 刘跃 | Method for data fragmentation of distributed storage system |
| CN104050102A (en) * | 2014-06-26 | 2014-09-17 | 北京思特奇信息技术股份有限公司 | Object storing method and device in telecommunication system |
| CN105991705A (en) * | 2015-02-10 | 2016-10-05 | 中兴通讯股份有限公司 | Distributed storage system and method of realizing hard affinity of resource |
| CN106055277A (en) * | 2016-05-31 | 2016-10-26 | 重庆大学 | Decentralized distributed heterogeneous storage system data distribution method |
| CN106339181A (en) * | 2016-08-19 | 2017-01-18 | 华为技术有限公司 | Method and system for processing data in storage system |
| CN106453611A (en) * | 2016-11-10 | 2017-02-22 | 郑州云海信息技术有限公司 | A method and apparatus for load balancing at a plurality of storage nodes |
| CN107066206A (en) * | 2017-03-22 | 2017-08-18 | 佛山科学技术学院 | The storage controlling method and system of a kind of distributed physical disk |
| CN107153513A (en) * | 2017-03-22 | 2017-09-12 | 佛山科学技术学院 | A kind of storage controlling method and server of distribution system services device |
| CN107168645A (en) * | 2017-03-22 | 2017-09-15 | 佛山科学技术学院 | The storage controlling method and system of a kind of distributed system |
| CN107577543A (en) * | 2017-09-18 | 2018-01-12 | 郑州云海信息技术有限公司 | Data read-write method, device, storage system and computer-readable recording medium |
| CN107948248A (en) * | 2017-11-01 | 2018-04-20 | 平安科技（深圳）有限公司 | Distributed storage method, control server and computer-readable recording medium |
| CN104754000B (en) * | 2013-12-30 | 2018-08-24 | 国家电网公司 | A kind of load-balancing method and system |
| WO2020010502A1 (en) * | 2018-07-10 | 2020-01-16 | 深圳花儿数据技术有限公司 | Distributed data redundant storage method based on consistent hash algorithm |
| CN111131457A (en) * | 2019-12-25 | 2020-05-08 | 上海交通大学 | Capacity and bandwidth compromise method and system for heterogeneous distributed storage |
| CN111210879A (en) * | 2020-01-06 | 2020-05-29 | 中国海洋大学 | Hierarchical storage optimization method for super-large-scale drug data |
| CN111245924A (en) * | 2020-01-08 | 2020-06-05 | 北京松果电子有限公司 | Load balancing method and device and computer storage medium |
| US11106488B2 (en) | 2018-08-01 | 2021-08-31 | Advanced New Technologies Co., Ltd. | Blockchain read/write data processing method, apparatus, and server |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521330A (en) * | 2011-12-07 | 2012-06-27 | 华中科技大学 | Mirror distributed storage method under desktop virtual environment |
| CN102638566A (en) * | 2012-02-28 | 2012-08-15 | 山东大学 | BLOG system running method based on cloud storage |
-
2013
- 2013-03-21 CN CN2013100917711A patent/CN103124299A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521330A (en) * | 2011-12-07 | 2012-06-27 | 华中科技大学 | Mirror distributed storage method under desktop virtual environment |
| CN102638566A (en) * | 2012-02-28 | 2012-08-15 | 山东大学 | BLOG system running method based on cloud storage |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104754000B (en) * | 2013-12-30 | 2018-08-24 | 国家电网公司 | A kind of load-balancing method and system |
| CN103929500A (en) * | 2014-05-06 | 2014-07-16 | 刘跃 | Method for data fragmentation of distributed storage system |
| CN104050102A (en) * | 2014-06-26 | 2014-09-17 | 北京思特奇信息技术股份有限公司 | Object storing method and device in telecommunication system |
| CN105991705A (en) * | 2015-02-10 | 2016-10-05 | 中兴通讯股份有限公司 | Distributed storage system and method of realizing hard affinity of resource |
| CN106055277A (en) * | 2016-05-31 | 2016-10-26 | 重庆大学 | Decentralized distributed heterogeneous storage system data distribution method |
| CN106339181A (en) * | 2016-08-19 | 2017-01-18 | 华为技术有限公司 | Method and system for processing data in storage system |
| CN106339181B (en) * | 2016-08-19 | 2019-05-24 | 华为技术有限公司 | Data processing method and device in storage system |
| CN106453611A (en) * | 2016-11-10 | 2017-02-22 | 郑州云海信息技术有限公司 | A method and apparatus for load balancing at a plurality of storage nodes |
| CN106453611B (en) * | 2016-11-10 | 2019-07-26 | 郑州云海信息技术有限公司 | A kind of method and device of more memory node load balancing |
| CN107066206A (en) * | 2017-03-22 | 2017-08-18 | 佛山科学技术学院 | The storage controlling method and system of a kind of distributed physical disk |
| CN107153513B (en) * | 2017-03-22 | 2020-07-24 | 佛山科学技术学院 | Storage control method of distributed system server and server |
| CN107168645B (en) * | 2017-03-22 | 2020-07-28 | 佛山科学技术学院 | Storage control method and system of distributed system |
| CN107168645A (en) * | 2017-03-22 | 2017-09-15 | 佛山科学技术学院 | The storage controlling method and system of a kind of distributed system |
| CN107153513A (en) * | 2017-03-22 | 2017-09-12 | 佛山科学技术学院 | A kind of storage controlling method and server of distribution system services device |
| CN107066206B (en) * | 2017-03-22 | 2020-07-24 | 佛山科学技术学院 | Storage control method and system for distributed physical disk |
| CN107577543A (en) * | 2017-09-18 | 2018-01-12 | 郑州云海信息技术有限公司 | Data read-write method, device, storage system and computer-readable recording medium |
| CN107948248A (en) * | 2017-11-01 | 2018-04-20 | 平安科技（深圳）有限公司 | Distributed storage method, control server and computer-readable recording medium |
| WO2020010502A1 (en) * | 2018-07-10 | 2020-01-16 | 深圳花儿数据技术有限公司 | Distributed data redundant storage method based on consistent hash algorithm |
| US11106488B2 (en) | 2018-08-01 | 2021-08-31 | Advanced New Technologies Co., Ltd. | Blockchain read/write data processing method, apparatus, and server |
| EP3779691A4 (en) * | 2018-08-01 | 2021-10-13 | Advanced New Technologies Co., Ltd. | Data processing method and apparatus, and server |
| CN111131457A (en) * | 2019-12-25 | 2020-05-08 | 上海交通大学 | Capacity and bandwidth compromise method and system for heterogeneous distributed storage |
| CN111131457B (en) * | 2019-12-25 | 2021-11-30 | 上海交通大学 | Capacity and bandwidth compromise method and system for heterogeneous distributed storage |
| CN111210879A (en) * | 2020-01-06 | 2020-05-29 | 中国海洋大学 | Hierarchical storage optimization method for super-large-scale drug data |
| CN111245924A (en) * | 2020-01-08 | 2020-06-05 | 北京松果电子有限公司 | Load balancing method and device and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103124299A (en) | Distributed block-level storage system in heterogeneous environment | |
| US11301144B2 (en) | Data storage system | |
| US11444641B2 (en) | Data storage system with enforced fencing | |
| US11438411B2 (en) | Data storage system with redundant internal networks | |
| US11467732B2 (en) | Data storage system with multiple durability levels | |
| US20220057951A1 (en) | Data storage system with metadata check-pointing | |
| US11237772B2 (en) | Data storage system with multi-tier control plane | |
| CN107707393B (en) | Multi-active system based on Openstack O version characteristics | |
| US11146626B2 (en) | Cloud computing environment with replication system configured to reduce latency of data read access | |
| US10057187B1 (en) | Dynamic resource creation to connect client resources in a distributed system | |
| CN109690494B (en) | Hierarchical fault tolerance in system storage | |
| US10616134B1 (en) | Prioritizing resource hosts for resource placement | |
| CN106993064A (en) | A kind of system and its construction method and application that the storage of mass data scalability is realized based on Openstack cloud platforms | |
| Zheng et al. | Design and research on private cloud computing architecture to support smart grid | |
| US10154091B1 (en) | Deploying infrastructure units according to resource hosting constraints | |
| US11182096B1 (en) | Data storage system with configurable durability | |
| US11405455B2 (en) | Elastic scaling in a storage network environment | |
| CN104410531A (en) | Redundant system architecture approach | |
| KR20230091168A (en) | Techniques for creating configurations to electrically isolate fault domains in data centers | |
| US11336519B1 (en) | Evaluating placement configurations for distributed resource placement | |
| KR101673882B1 (en) | Storage system with virtualization using embedded disk and method of operation thereof | |
| Salapura et al. | Enabling enterprise-level workloads in the enterprise-class cloud | |
| CN110764692B (en) | Method and system for redundancy and recovery of storage slice data in server cluster environment | |
| Lin et al. | A job-oriented load-distribution scheme for cost-effective NameNode service in HDFS | |
| Shan et al. | Explore Data Placement Algorithm for Balanced Recovery Load Distribution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130529 |