CN105450741B - Provisioning server selection and repair data transmission method in distributed memory system - Google Patents
Provisioning server selection and repair data transmission method in distributed memory system Download PDFInfo
- Publication number
- CN105450741B CN105450741B CN201510784810.5A CN201510784810A CN105450741B CN 105450741 B CN105450741 B CN 105450741B CN 201510784810 A CN201510784810 A CN 201510784810A CN 105450741 B CN105450741 B CN 105450741B
- Authority
- CN
- China
- Prior art keywords
- node
- supply
- newborn
- data
- selection
- 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
-
- 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 discloses provisioning server selection and repair data transmission methods in a kind of distributed memory system, it is characterized in that, it is realized based on mixed integer linear programming, provisioning server is defined as supply node, according to distributed memory system to be repaired, obtain node set, link set, need selection supply node quantity, each supply the data volume that node need to be transferred to newborn node；Distributed memory system based on network code is expressed as digraph；According to digraph, it is based on mixed integer linear programming, establishes the mathematical model of optimal supply node selection and repair data transmission；It solves and obtains optimal supply node set, it is each to supply the message transmission rate transmitted on node to the data transfer path of newborn node and each path.The present invention reduces the recovery times of entire data reparation, to improve system performance.
Description
Technical field
The present invention relates to the data reparations of the distributed memory system based on network code, and in particular to one kind is to being based on
The distributed memory system of network code carries out the method that provisioning server selection and repair data transmission are carried out when data reparation.
Background technology
With the arrival in big data epoch, the information of explosive growth accelerates the development of data storage.Cisco's prediction exists
2018, data center's flow was estimated as 8.6 ZB（zettabytes）.Since distributed memory system has the following advantages：1）
Enhance the reliability of storage；2）Improve the reliability of storage；3）Dynamic scalable, for ease of maintenaince etc., therefore, much provide number
It is that data are stored by distributed memory system according to the service commercial city of storage.
It is well known that the storage of data must have very high reliability.In order to ensure the reliability and section of data storage
About cost, in the prior art, large-scale distributed storage system are led to generally on being distributed in the server of diverse geographic location
Redundant fashion storage data are crossed, in this way, user can obtain required information anywhere or anytime.Traditional storage scheme mainly has
Duplication and Reed Solomon code.2000, network code was suggested, and network code is a kind of letter having merged routing and coding
Cease switching technology, its core concept be on each node in a network to the information received on each channel carry out it is linear or
The nonlinear processing of person, is then forwarded to downstream node, and intermediate node plays the role of encoder or signal processor.Pass through
Network code can reach max-flow circle of QoS routing transmission, improve the efficiency of transmission of information, thus it has increase net
Network handling capacity, improves the advantages that bandwidth availability ratio at balance network load.Network code is applied to the data of distributed memory system
Storage, while there is the MDS properties of correcting and eleting codes, as long as user obtains enough encoding blocks and restores original.
Although network code can improve reliability, in practical applications, data in distributed memory system be easy by
It to miscellaneous destruction, such as powers off, hacker attacks, disk failures etc..In order to ensure the reliability of storage, need to losing
Data repaired in time.After some storage server fails, a new storage server（The server is referred to as
Emergent services device）Original failed server can be replaced, and from the storage server of other survivals（It also referred to as survives server）
Middle selected section server（Referred to as provisioning server）, and from repair data is downloaded thereon, the data of loss are regenerated.Example
Such as, Chinese invention patent application CN103336785A discloses a kind of distributed storage method and its dress based on network code
It sets, wherein in data repair step, from data available node（Survival server）Middle random selection coded data block is downloaded
Coded data block structure, which is repaired matrix and stored, arrives new data available node（Emergent services device）In.
But in repair process, the selection of provisioning server and the transmission method of repair data directly affect network
Bandwidth（Repair bandwidth）Consumption and reparation time delay（Recovery time）.The method of above-mentioned random selection provisioning server is to the recovery time
It has adverse effect.
Invention content
The goal of the invention of the present invention is to provide provisioning server selection and repair data in a kind of distributed memory system and passes
Transmission method efficiently uses network bandwidth by the optimization of optimum choice and data transfer path to provisioning server, and shortening is repaiied
Multiple time delay.
To achieve the above object of the invention, the technical solution adopted by the present invention is：Clothes are supplied in a kind of distributed memory system
The selection of business device and repair data transmission method, are realized based on mixed integer linear programming, and provisioning server is defined as to supply section
Point, the described method comprises the following steps：
A, according to distributed memory system to be repaired, following parameters are obtained：
The node set being made of the InterWorking Equipment in survival server, emergent services device and network,
Link set in node set between each node,
The quantity of the supply node of selection is needed,
Each supply node need to be transferred to the data volume of newborn node；
B, according to acquired parameter, the distributed memory system based on network code is expressed as digraph G (V, E), G
For directed connected graph, V and E are respectively node set and link set；
Wherein, it is newborn node with emergent services device, with the interconnection in network using the survival server as surviving node
Equipment is intermediate node, and the node set is made of surviving node, newborn node and intermediate node；
The construction method of the link set is, if there are communication link between two different nodes in node set V,
A link is then connected between the two nodes, and is provided in the data transfer direction on communication link and each transmission direction
Bandwidth, repeat this step until all qualified link connections finish, these links constitute link set E；
C, according to digraph, it is based on mixed integer linear programming, establishes optimal supply node selection and repair data transmission
Mathematical model；
D, it for the mathematical model of foundation, solves and obtains optimal supply node set, each node of supplying is to new tight knot
The message transmission rate transmitted on the data transfer path and each path of point；
E：Obtain the scheme of optimal supply node selection and repair data transmission.
In above-mentioned technical proposal, the step C specifically includes following steps：
C1, known parameters are obtained：The gathering of surviving node, newborn node, intermediate node, each node upstream neighbor section
Point set, the downstream neighboring node set of each node, the set of directed edge, the bandwidth on directed edge, the supply section for needing selection
The quantity of point and each supply node need to be transferred to the data volume of newborn node；
C2, variable is defined, the variable is：Each supply node-node transmission to newborn node streaming rate and
The biography of the minimum value of the rate, the message transmission rate per data stream in each edge, each supply node to newborn node
Whether defeated rate and current surviving node are chosen as supply node, and 0 indicates that current surviving node is not chosen as supply node, and 1
Indicate that current surviving node is chosen as supply node；
C3, object function is built using the parameter and variable of the definition --- the time for minimizing data reparation is equivalent to
Maximize the minimum value of streaming rate of all supply node-node transmissions to newborn node：
Building bound for objective function includes：It is all supply node-node transmissions to newborn node streaming rate most
Small value is less than or equal to each supply node-node transmission to the transmission rate of newborn node, the data transmission speed of each intermediate node of disengaging
Message transmission rate conservation, the different numbers received in the message transmission rate and new life node that rate conservation, supply node are sent out
The bandwidth of the link is no more than by total rate of same link according to stream, ensures only have surviving node to be selected as supplying node
When, there is data flow from node outflow, the value range of the quantity of supply on restriction node and integer variable and real variable；
Thus the mathematical model of optimal supply node selection and repair data transmission is established.
Above, it is mainly in view of following factor when building constraints：Data repair time is each supply node
It will need the maximum value of the data whole end of transmission time transmitted.For each supply node, it would be desirable to which the data of transmission are complete
Portion's end of transmission time is equal to the data volume that need to be transmitted divided by each supply node to the transmission rate of newborn node.Because in base
In the distributed memory system of network code, each it is chosen as supply node-node transmission and arrives the data volume of newborn node to be identical
Value, so, minimize data repair time, be equivalent to maximize each supply node to the transmission rate of new life node minimum
Value.
In above-mentioned technical proposal, integral linear programming method for solving is implicit enumeration method, branch and bound method or cutting plane algorithm.
Further technical solution using AMPL Language Modelings and uses CPLEX to the mixed integer linear programming that is proposed
Linear programming for solution device direct solution obtains provisioning server selection and repair data transmission plan.
Since above-mentioned technical proposal is used, the present invention has following advantages compared with prior art：
Different from random selection in the prior art, distributed system is converted to digraph by the present invention, and supply is serviced
The selection of device and data transmission scheme is converted into optimization problem, and optimum choice is realized by mixed integer linear programming, optimal
Corresponding provisioning server set is selected in ground from survival server set, and obtains from provisioning server to emergent services device
Between data transmission scheme, reduce the recovery time of entire data reparation, to improve system performance.
Description of the drawings
Fig. 1 is optimal supply node sum number in the repair process of the distributed memory system based on network code in embodiment
According to the flow chart of transmission plan；
Fig. 2 is the network model of the distributed memory system based on network code；
Fig. 3 is the digraph of network model structure according to fig. 2；
Fig. 4-Fig. 6 is three kinds of different recovery scenario schematic diagrames；
Fig. 7 is simulation result schematic diagram in embodiment.
Specific implementation mode
The invention will be further described with reference to the accompanying drawings and embodiments：
Embodiment：Present embodiments provide it is a kind of based on network code data storage interior joint lose repair process in most
The method of excellent supply node selection and data transmission, it is contemplated that node is supplied in the distributed memory system based on network code
Selection and data transfer path so that the reparation minimizing delay in data repair process.
Fig. 1 be the present embodiment the distributed memory system based on network code repair process in optimal supply node with
The flow chart of data transmission scheme, is as follows：
A11：Obtain the distributed memory system parameter based on network code.
It is specific as follows：
According to the distributed memory system based on network code, parameter below is obtained：Node set（It is saved including survival
The intermediate node and newborn node that point, router/network interconnection apparatus, network interconnection are constituted）, inter-node link link information, link bandwidth
（Link maximum transmission rate）, need selection supply node quantity and each supply node need to be transferred to the number of newborn node
According to amount.
It needs the supply number of nodes of selection to be arranged to be the arbitrary integer more than or equal to 1, according to actual conditions by system
Or manager is given.In general, can restore original document from arbitrary k surviving node download data, then can be arranged needs to select
The supply number of nodes selected is k.
A12：According to the parameter obtained in step A11, the distributed memory system based on network code is expressed as oriented
Figure,。
Wherein, V indicates digraphNode set, node set includes：Surviving node set, newborn node and in
Segmentum intercalaris point set, E indicate the link set of digraph G.
Fig. 2 gives the reparation example of the distributed memory system based on network code.From Fig. 2, it will be seen that network node
Including 4 surviving nodes（,）, 1 newborn node（）, during 7 routers/network interconnection apparatus, network interconnection constitutes
Intermediate node（）, repaired wherein needing to select 3 from 4 surviving nodes and participating in data as supply nodes
Cheng Zhong.The maximum data rate of link between neighbouring device is as shown in the figure.
A121：Distributed memory system based on network code is expressed as digraph, is as follows：
In physical network, judge to whether there is communication link in network between two equipment, if it is, in digraph
In a link is connected between corresponding two nodes, and the data transmission on digraph link is provided according to communication link information
Otherwise direction is not connected to any link between the two nodes.All links are indicated with link set E.For example, it is assumed that section
There are a communication links from u to v between point u and node v, then there is a directed link from u to v between them,.The peak transfer rate of each of the links is obtained according to the bandwidth situation of communication link in physical network, is indicated
For。
A121 repeat the above steps until having traversed all qualified links.
Surviving in the example that Fig. 2 is provided, in network, server, router and emergent services device are total 12, phase
It answers, it is the digraph built according to the example of Fig. 2 to have 12 nodes, Fig. 3 in digraph.
Therefore, Fig. 3 is the digraph of example structure according to fig. 2, and Fig. 4, Fig. 5, Fig. 6 give different supplies
Node selects and repair data transmission plan.Entire file size is M=300Mb,=2, supply number of nodes=3, each
Storage server storage data volume be=150Mb, therefore the data volume of each supply node-node transmission is=75Mb.The supply node that Fig. 4 is selected for,, wherein from supply node to new tight knot
Three paths of point all pass through link, therefore the maximum data transmission rate of each path on the roads Tiao Lian is
30Mbps/3=10Mbps.The minimum value for all chain road the maximum data rates that each path is passed through is 10 Mbps.Therefore it
Recovery time be 75 Mb/10Mbps=7.5s；The supply node that Fig. 5 is selected for,, wherein having two paths all
Pass through link, therefore the maximum data transmission rate of each path on the roads Tiao Lian is 30/2=15Mbps.Every road
The minimum value for all chain road the maximum data rates that diameter passes through is 15 Mbps, therefore its recovery time is 75 Mb/15
Mbps =5s；Fig. 6 selection supply node be also,, as shown, You Santiao road different from the path of Fig. 5 selections
Diameter all passes through link, its maximum data rate is 100 Mbps, therefore each path is in the maximum data of this link
Transmission rate is 100/3=33.33Mbp.The minimum value for all chain road the maximum data rates that each path is passed through is 30
Mbps, therefore the maximum data rate of each path transmitting data is 30Mbps.Fig. 6 recovery times be the Mbps of 75 Mb/30=
2.5s, it is seen that influence of the scheme of the selection of supply node and data transmission to the recovery time.
A13：According to digraph, build the selection of optimal supply node and data transmission based on mixed integer linear programming
Mathematical model.
Linear programming is that object function and constraints are all linear optimization problems, is largely used to solve maximization
Or the practical problem of minimization.The mixed integer linear programming of the present invention is one kind of linear programming.
According to digraph, build the number of the selection of optimal supply node and data transmission based on mixed integer linear programming
Model is learned, is as follows：
A131：Known parameters are defined.
In the present embodiment, parameter below is defined：Node setV,VIncluding surviving node set,, in formula,n i Indicate theiA surviving node, newborn nodeWith network interconnection node set, directed edge setE, from nodeuIt is directed toward nodevDirected edge,, link bandwidth/the maximum data rate
, each node upstream and downstream node set, need selection supply node quantitykAnd each supply
Node needs the data volume for being transferred to newborn node, surviving node indexed set, one it is sufficiently large
Constant；
A132：Known variables are defined.
In the present embodiment, parameter below is defined：Each supply node to newborn node transmission rate
Minimum value（The system regeneration time is）, in digraphGIn, different surviving nodes() to newborn node
Transmission rate (), streamOn sideOn flowAnd 0-1 variables, work as survival
NodeIt is selected as supplying node,, otherwise。o i To constrain the recovery timerNeeds variable, be used for
Target is set to obtainrValue is
A133：Build the mathematical model of optimal supply node selection and data transmission based on mixed integer linear programming.
The optimization object of the present invention is the time of data reproduction.Data reproduction timing definition, which will be each supply node, to be needed
The maximum value of the data whole end of transmission time of transmission.For each supply node, it would be desirable to which the data of transmission are all transmitted
The time finished is equal to transmitted data amount divided by each supply node to the transmission rate of newborn node.Because being compiled based on network
In the distributed memory system of code, each it is chosen as supply node-node transmission and arrives the data volume of newborn node to be identical value, so, most
Smallization data repair time (), whereinrEach supply node is indicated to the minimum value of the transmission rate of newborn node, etc.
Valence is in the minimum value of each supply node of maximization to the transmission rate of newborn node（rObject as an optimization）.
A1331：Structure integer model first is as follows：
Object function：（1）
Formula (1) is object function, maximizes the rate of data transmission, that is, maximizes supply node to the number of newborn node
According to the minimum value of transmission rate.
Constraints is：
Formula (2)-(11) are constraints, the selection of supply on restriction node, selectionA surviving node is as supply node
Keep the data reproduction time minimum, that is, maximizes supply node to the minimum value of the message transmission rate of newborn node.Formula（2）（3）
InIt is set as a sufficiently large constant.Formula（2）（3）Constraint makesFor the minimum value of each supply node data transmission rate.
Work as surviving nodeWhen being chosen as supply node,=1, then, otherwise,；Formula（4）-（6）For
Ensure that the data transmission of whole network meets stream law of conservation；Formula（4）-（6）With（8）Ensure to be chosen as supply section when surviving node
When point, there are non-zero transmission rates, that is, there are data and reach newborn node from the node；Formula（7）Define different circulations
Cross the limitation of the overall transmission rate of same link；Formula (9) ensure that and can only selectA surviving node is as supply node；
Formula (10)-(11) give the value range of 0-1 integer variables and real variable.
A14：For the mathematical model of foundation, by the linear programming in the AMPL language description present invention and CPLEX is used
Solver solves, and obtains optimal solution, further obtains optimal supply node set and repair data transmission plan.
In the present embodiment, using CPLEX linear programming for solution device direct solutions.If according to solver obtain as a result,Value be 0, show nodeIt is not chosen as supply node；If value is 1, show nodeIt is chosen as supply node.VariableValue indicate from nodeThe repair data of newborn node is dealt into linkOn data
Transmission rate.If nodeIt is not chosen as supply node, i.e.,=0, then solution obtains.Become
AmountValue constitute repair data transmission plan.It is selected according to the supply node of result
Keep the data reproduction time minimum with repair data transmission plan.
Further to verify the effect of the present invention, in emulation experiment, generated using the random network topology being widely recognized as
Algorithm Waxman algorithms generate the distributed memory system based on network code network topology, the algorithm include three parameter alphas,
β, γ, wherein α control the node density in network topology, and there are the probability of link between β and γ control nodes.The node quilt of generation
Random is arranged inRegion in.The bandwidth of each of the links is set as being evenly distributed on [10Mbps, 40Mbps]
In range；The quantity of selection supply nodeRanging from 4 to 9.In emulation experiment, the file of storage passes throughMDS redundancy encodings are stored in server（Surviving node）On.File size is M=1024Mb, each to take
The data volume stored on business device, each data volume for supplying node-node transmission is, testing
In, as known quantity.
In the emulation experiment of the present invention, α=0.4 is set, when β=0.4, γ=0.4, a topology is generated, is then directed to this
Topological diagram is opened, the recovery time is obtained according to proposed algorithm.Table 1 and Fig. 7 are simulation result, and wherein table 1 is real using the present invention
The method for applying example offer obtains the data reproduction time compared with the recovery time that random selection supply nodal method obtains, Fig. 4 handles
Simulation result in table 2 has carried out intuitive displaying by line chart.
Table 1
| Supply the quantity of node | Recovery time under random selection scheme（s） | Recovery time under scheme of the present invention（s） |
| 4 | 21.5 | 14.9 |
| 5 | 18.9 | 11.3 |
| 6 | 13.8 | 9.4 |
| 7 | 10.3 | 8.7 |
| 8 | 9.1 | 7.9 |
| 9 | 8.4 | 6.1 |
In Fig. 7, it is shown that the simulation result diagram that method provided in an embodiment of the present invention is emulated（Supply number of nodes
Variation）.As can be seen that when although d values increase from table 1 and Fig. 7, the recovery time is all being reduced under two schemes, this
Recovery time of the scheme recovery time of invention significantly lower than random selection supply node.
A15：Obtain the scheme of optimal supply node set and data transmission.
Method provided in an embodiment of the present invention considers in the distributed system based on network code, when there is storage service
When device fails and is replaced with emergent services device, by from others survive servers in, select a certain number of server sets for
Newborn server transmission data, to reduce the recovery time of data.
In addition, provisioning server collection in network and data transmission problems are used the linear programming method in mathematics by the present invention
It solves, by, using AMPL Language Modelings and with CPLEX linear programming for solution device direct solutions, being obtained to mixed integer linear programming
Take optimal solution.Method provided by the invention has good application value, for those offer distributed storage data services
Big companies（Such as google）, the damage or failure of server frequently occur, using the present invention, can greatly reduce and repair
The multiple time reduces the time delay for interrupting service user, can improve system performance to a certain extent.
Claims (3)
1. provisioning server selection and repair data transmission method in a kind of distributed memory system, which is characterized in that based on mixed
It closes integral linear programming to realize, provisioning server is defined as supply node, the described method comprises the following steps：
A, according to distributed memory system to be repaired, following parameters are obtained：
The node set being made of the InterWorking Equipment in survival server, emergent services device and network,
Link set in node set between each node,
The quantity of the supply node of selection is needed,
Each supply node need to be transferred to the data volume of newborn node；
B, according to acquired parameter, the distributed memory system based on network code is expressed as digraph G (V, E), G is to have
To connected graph, V and E are respectively node set and link set；
Wherein, it is newborn node with emergent services device, with the InterWorking Equipment in network using the survival server as surviving node
For intermediate node, the node set is made of surviving node, newborn node and intermediate node；
The construction method of the link set is, if there are communication link between two different nodes in node set V,
A link is connected between two nodes, and provides the data transfer direction on communication link and the band in each transmission direction
Width repeats this step until all qualified link connections finish, these links constitute link set E；
C, according to digraph, it is based on mixed integer linear programming, establishes the number of optimal supply node selection and repair data transmission
Learn model；
D, it for the mathematical model of foundation, solves and obtains optimal supply node set, each node of supplying is to newborn node
The message transmission rate transmitted on data transfer path and each path；
E：Obtain the scheme of optimal supply node selection and repair data transmission；
Wherein, the step C specifically includes following steps：
C1, known parameters are obtained：The gathering of surviving node, newborn node, intermediate node, each node upstream neighbor set of node
Conjunction, the downstream neighboring node set of each node, the set of directed edge, the bandwidth on directed edge, the supply node for needing selection
Quantity and each supply node need to be transferred to the data volume of newborn node；
C2, variable is defined, the variable is：Each node-node transmission of supplying is to the streaming rate and the speed of newborn node
The minimum value of rate, the message transmission rate per data stream in each edge, the transmission speed of each supply node to newborn node
Whether rate and current surviving node are chosen as supply node, and 0 indicates that current surviving node is not chosen as supply node, and 1 indicates
Current surviving node is chosen as supply node；
C3, object function is built using the parameter and variable of the definition --- it the time for minimizing data reparation, is equivalent to most
Change the minimum value of streaming rate of all supply node-node transmissions to newborn node greatly：
Building bound for objective function includes：The minimum value of streaming rate of all supply node-node transmissions to newborn node
It is kept less than or equal to the transmission rate of each supply node-node transmission to newborn node, the message transmission rate of each intermediate node of disengaging
Perseverance supplies the message transmission rate conservation received in the message transmission rate and newborn node that node is sent out, different data flows
When being no more than the bandwidth of the link by total rate of same link, ensure to only have surviving node to be selected as supplying node,
There is data flow from node outflow, the value range of the quantity of supply on restriction node and integer variable and real variable；Thus
Establish the mathematical model of optimal supply node selection and repair data transmission.
2. provisioning server selection and repair data transmission side in a kind of distributed memory system according to claim 1
Method, it is characterised in that：Integral linear programming method for solving is implicit enumeration method, branch and bound method or cutting plane algorithm.
3. provisioning server selection and repair data transmission side in a kind of distributed memory system according to claim 2
Method, it is characterised in that：To the mixed integer linear programming that is proposed using AMPL Language Modelings and with CPLEX linear programming for solution
Device direct solution obtains provisioning server selection and repair data transmission plan.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510784810.5A CN105450741B (en) | 2015-11-16 | 2015-11-16 | Provisioning server selection and repair data transmission method in distributed memory system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510784810.5A CN105450741B (en) | 2015-11-16 | 2015-11-16 | Provisioning server selection and repair data transmission method in distributed memory system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105450741A CN105450741A (en) | 2016-03-30 |
| CN105450741B true CN105450741B (en) | 2018-09-25 |
Family
ID=55560507
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510784810.5A Active CN105450741B (en) | 2015-11-16 | 2015-11-16 | Provisioning server selection and repair data transmission method in distributed memory system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105450741B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106201356B (en) * | 2016-07-14 | 2019-07-19 | 北京理工大学 | A kind of dynamic data dispatching method based on link available bandwidth state |
| CN106304244B (en) * | 2016-08-15 | 2019-06-07 | 北京邮电大学 | A kind of the low energy consumption transmission method and device of multi-hop cellular network |
| CN107770077B (en) * | 2016-08-23 | 2021-01-29 | 苏州大学 | Information theory security multicast routing selection method based on network coding |
| CN108462643B (en) * | 2016-12-11 | 2021-02-02 | 苏州大学 | Integer transmission rate-oriented weak security multicast transmission topology construction method based on network coding |
| EP3605340B1 (en) * | 2017-05-31 | 2021-03-31 | Mitsubishi Electric Corporation | Data replicator and data replication program |
| CN110516385B (en) * | 2019-08-30 | 2023-04-25 | 长沙理工大学 | Method, system and related device for establishing data transmission network |
| CN111245719B (en) * | 2020-01-02 | 2021-12-31 | 湖南大学 | Ant colony optimization-based erasure coding storage system data updating method |
| CN112118604B (en) * | 2020-07-27 | 2023-06-20 | 哈尔滨工业大学（深圳） | Mobile storage system-oriented relay cooperation data restoration method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102546025A (en) * | 2012-01-11 | 2012-07-04 | 苏州大学 | Optimal distribution method for optical network units in optical fiber wireless hybrid access network |
| CN104660663A (en) * | 2013-11-19 | 2015-05-27 | 群晖科技股份有限公司 | Operation method of server cluster |
-
2015
- 2015-11-16 CN CN201510784810.5A patent/CN105450741B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102546025A (en) * | 2012-01-11 | 2012-07-04 | 苏州大学 | Optimal distribution method for optical network units in optical fiber wireless hybrid access network |
| CN104660663A (en) * | 2013-11-19 | 2015-05-27 | 群晖科技股份有限公司 | Operation method of server cluster |
Non-Patent Citations (3)
| Title |
|---|
| Optimal Node Selection for Data Regeneration in Heterogeneous Distributed Storage Systems;Gong, Q;《2015 44th International Conference on Parallel Processing》;20150930;第390-399页 * |
| Topology-Aware Node Selection for Data Regeneration in Heterogeneous Distributed Storage Systems;Gong, Qingyuan; Wang, Jiaqi; Wang, Yan; Wei, Dongsheng; Wang, Ji;《arXiv e-print,Bibliographic Code:2015arXiv150605579G》;20150630;第1-14页 * |
| 分布式存储系统中节点修复问题研究;张凯;《中国优秀硕士学位论文全文数据库（电子期刊）》;20121015;第I137-33页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105450741A (en) | 2016-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105450741B (en) | Provisioning server selection and repair data transmission method in distributed memory system | |
| Chiu et al. | Can network coding help in P2P networks? | |
| Ho et al. | Network coding: an introduction | |
| CN103348639B (en) | Routing resource and Control Server | |
| CN105830377A (en) | Dynamic coding for network traffic by fog computing node | |
| Gerami et al. | Optimal-cost repair in multi-hop distributed storage systems | |
| CN105791169A (en) | Switch transmission control method, switch transmitting method and related equipment in SDN (Software Defined Network) | |
| CN106549779A (en) | A kind of maximum non-intersect double route collocation method of multiple constraint energy communication service | |
| CN103781115B (en) | Distributed base station buffer replacing method based on transmission cost in a kind of cellular network | |
| Milosavljevic et al. | Efficient algorithms for the data exchange problem | |
| CN104601473A (en) | Route generating method and system for multi-target path with restriction | |
| CN102447640B (en) | Method for carrying out aggregate routing in IRF (intelligent resilient framework) system and machine frame switch | |
| CN106104503B (en) | Using unified API to for forwarding server and framework program to be used for the fine granularity network optimization | |
| Kim et al. | An evolutionary approach to inter-session network coding | |
| CN111865789B (en) | SR path constraint method based on segment routing | |
| ParandehGheibi et al. | Optimal reverse carpooling over wireless networks-a distributed optimization approach | |
| CN102325009A (en) | Forward error correction-based method for reliably transmitting network coding multicast data stream | |
| Pan et al. | GOR: Group-oblivious multicast routing in airborne tactical networks under uncertainty | |
| CN107872385A (en) | A kind of SDN router-level topology and control method | |
| Zhang et al. | Repair tree: fast repair for single failure in erasure-coded distributed storage systems | |
| CN104917677A (en) | Data stream forwarding control method and system | |
| Pašić et al. | Minimum cost survivable routing algorithms for generalized diversity coding | |
| Guo et al. | Codedstream: Live media streaming with overlay coded multicast | |
| CN109327407A (en) | DEU data exchange unit, method for interchanging data, computer equipment and storage medium | |
| CN108632066A (en) | A kind of method and apparatus of structure Video Multicast virtual net |
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 | ||
| GR01 | Patent grant |