CN107508906A - Distributed file system node selecting method and master server - Google Patents
Distributed file system node selecting method and master server Download PDFInfo
- Publication number
- CN107508906A CN107508906A CN201710821141.3A CN201710821141A CN107508906A CN 107508906 A CN107508906 A CN 107508906A CN 201710821141 A CN201710821141 A CN 201710821141A CN 107508906 A CN107508906 A CN 107508906A
- Authority
- CN
- China
- Prior art keywords
- user
- node
- period
- master server
- zone
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- 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
-
- 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
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1029—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers using data related to the state of servers by a load balancer
-
- 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
- H04L67/1095—Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Information Transfer Between Computers (AREA)
- Computer And Data Communications (AREA)
Abstract
The present invention relates to a kind of distributed file system node selecting method and server,Belong to distributed computing technology field,The distributed file system node selecting method and server pass through the desired calculating to data transfer average speed between division period time and node cluster and zone user,It can determine the currently optimal node cluster of data transmission quality between user to be responded,So as to realize request instruction that the node server in the node cluster selects a node under it to respond user to be responded,Efficient data, services are provided for user to be responded,The distributed file system node selecting method and master server cause distributed file system to randomly choose mode by traditional node and be changed into the dynamic node selection mode changed with the change of network environment,Improve the data transmission efficiency between distributed file system and user,It ensure that the data interaction quality between system and user,Further increase the service quality of distributed file system.
Description
Technical field
The present invention relates to distributed computing technology field, more particularly to a kind of distributed file system node selecting method
And master server.
Background technology
The network model of Large-scale Distributed Virtual Environment is broadly divided into three kinds：Distributed, centralization and hierachical structure.
Distributed network model may also be referred to as being node self-adapting formula network model, not have node server in the network system, by
Local node is responsible for the storage and network communication of data, such as the use such as NPSNET, DIVE is exactly self-adapting type structure, and it is led
It is that ensure that the real-time of whole network system to want advantage, and shortcoming is one of the system update when virtual environment changes or expands
Cause property will be difficult to ensure that；, there is node server in centralized network model, task is sent to node service by each node
Device, distributed unitedly after being calculated by node server, its major advantage is that solve the problems, such as system conformance and security, and it is lacked
Point is with the expansion of virtual environment network size, and the resource-constrained of server may result in the delay of server process speed；
Layer-stepping network model has merged the advantages of self-adapting type network model and centralized network model, is by multiple node servers
Global nodes offer service, node is connected with node server using centralization, and node server is used with master server and divided
Layered structure, this layer-stepping network model carry out subregion, each node server management point according to the quantity of node server
Node in area.Although the distributed file system based on layer-stepping network model can ensure the load balance of internal system,
But due to the randomness that user accesses, different access locations and different network environments are all by between the system of influence and user
Communication quality, therefore under dynamic network, the data transmission quality between system and user is unstable, between system and user
Data interaction quality can not be guaranteed, and have impact on the service quality of distributed file system.
The content of the invention
Based on this, it is necessary to for data of the existing distributed file system under dynamic network between system and user
The problem of transmission quality is unstable, there is provided a kind of distributed file system node selecting method and master server.
To solve the above problems, the present invention takes following technical scheme：
A kind of distributed file system node selecting method, comprises the following steps：
Obtain each described zone user in several zone users respectively with each in several node clusters
Data transfer average speed of the node cluster within D × F period, the D is default collection number of days and presses every day
According to default division rule carry out the period averagely divide, the F be to any one day carry out the period averagely divide and according to
Time sequencing division obtained period is numbered after numbering maximum；
By any one of zone user respectively with node cluster each described within the preset numbers period in each day
Data transfer average speed include described in a zone user and one as a set, each described set
The data transfer average speed of the node cluster within the preset numbers period in each day, and travel through all regions
User；
Polynomial curve-fit is carried out to each the described set obtained after the traversal all zone user and calculated
It is expected corresponding to each described set；
It will it is expected it is expected set as one corresponding to zone user described in identical, and in each expectation set
Node cluster corresponding to zone user corresponding to middle determination greatest hope and the greatest hope and the greatest hope, will be each
Zone user corresponding to the greatest hope, node cluster and the preset numbers period one corresponding to each greatest hope
One is correspondingly recorded in master server table；
When receiving when the request instruction that response user sends, according to the request instruction in the master server table
Searched and matched, please when find the target area user to match with the user to be responded and the request instruction
When seeking time is within the preset numbers period, the node server hair into node cluster corresponding to the target area user
Assignment instructions are sent, so that the node server selects interdependent node to carry out the user to be responded according to the assignment instructions
Response.
Correspondingly, the present invention also proposes a kind of master server for the selection of distributed file system node, master server
It is configured to perform procedure below：
The master server obtain each described zone user in several zone users respectively with several nodes
Data transfer average speed of each described node cluster within D × F period in cluster, the D are default collection number of days
And every day averagely divides according to the default division rule progress period, the F is that any one day progress period was put down
Division and sequentially in time to divide obtained period be numbered after numbering maximum；
The master server is by any one of zone user respectively with node cluster each described in the default of each day
As a set, each described set includes a region and used data transfer average speed in the numbering period
Family and the data transfer average speed of the node cluster within the preset numbers period in each day, and travel through
All zone users；
The master server carries out multinomial letter to each the described set obtained after the traversal all zone user
It is expected corresponding to each described set of number plan joint account；
The master server will it is expected as one it is expected to gather corresponding to zone user described in identical, and at each
It is described it is expected to determine in set corresponding to zone user corresponding to greatest hope and the greatest hope and the greatest hope
Node cluster, by node cluster corresponding to zone user corresponding to each greatest hope, each greatest hope and described pre-
If numbering period one-to-one corresponding is recorded in master server table；
When the master server is received when the request instruction that user sends is responded, the master server is asked according to described
Instruction is asked to be searched and matched in the master server table, when the master server is found and user's phase to be responded
The target area user of matching and when the request time of the request instruction is within the preset numbers period, the main service
Node server in node cluster corresponding to from device to the target area user sends assignment instructions, so that the node server
Interdependent node is selected to respond the user to be responded according to the assignment instructions.
To overcome user in distributed virtual environment to access caused by randomness to data interaction between system and user
Influence, the present invention propose above-mentioned distributed file system node selecting method and master server, the distributed field system
System node selecting method and master server are put down by dividing data transfer between period and node cluster and zone user to the time
The desired calculating of equal speed, can determine the currently optimal node cluster of data transmission quality between user to be responded, so as to
The request instruction that the node server in the node cluster selects a node under it to respond user to be responded is realized, to wait to respond
User provides efficient data, services.Distributed file system node selecting method proposed by the invention and master server make
Distributed file system randomly chooses mode by traditional node and is changed into the dynamic section changed with the change of network environment
Point selection mode, the data transmission efficiency between distributed file system and user is improved, ensure that between system and user
Data interaction quality, further increase the service quality of distributed file system.
Brief description of the drawings
Fig. 1 is the structural representation of distributed file system；
Fig. 2 is the schematic diagram of zone user in distributed file system；
Fig. 3 is the schematic flow sheet of distributed file system node selecting method in the one of embodiment of the present invention；
Fig. 4 is the curve matching figure obtained after emulation.
Embodiment
Technical scheme is described in detail below in conjunction with accompanying drawing and preferred embodiment.
Distributed file system (Distributed File System) refers to the physical memory resources of file system management
Not necessarily it is directly connected on the local node, but is connected by computer network with node.The design of distributed file system
Based on Client/Server pattern, typically it is made up of master server layer, node server layer and node layer, as shown in figure 1, with
Family end sends request (including file name, user profile etc.) to master server, and master server makes answer (transmission information) simultaneously
Corresponding task is recorded in task list, node server obtains Master Server Tasks and selects corresponding node response user terminal
Request.
Wherein, master server (master)：First layer in distributed file system, receive user and ask generation to be appointed
Business list and abnormality processing；
Node server (node server)：The request for the node response user being responsible under selection node cluster；
Node (node)：It is the minimum unit in distributed file system, for the request real-time response to user, deposits
Storage and calculating data.
Node cluster (node cluster)：In whole distributed file system, each node server is corresponding thereto
The LAN cluster definition that is formed of node be node cluster；
Zone user (region client)：For whole distributed file system user divided after, each piece
The general designation of all users in zoning, as shown in Fig. 2 after the user to whole distributed file system divides, is obtained
To zone user 1, region amounts to n zone user for 2 ..., zone user n.
Distributed file system node selecting method proposed by the invention and the main purpose of master server are to find one
It is individual to provide the user optimal service with the optimal node of user network JA(junction ambient), thus distributed file system and user it
Between network transmission quality be need consider an important factor for one of.The present invention is by a certain network segment of record on node cluster
Network average speed (or data transfer average speed) simultaneously calculates expectation using Function Fitting, while it is expected relatively and sieved
Relatively good node cluster is selected, corresponding node is found out on the node cluster selected and carries out data interaction with user, so as to protect
The quality of data transfer between the node in distributed file system and user is demonstrate,proved.
The present invention one of embodiment in, as shown in figure 3, distributed file system node selecting method include with
Lower step：
S100 obtain each zone user in several zone users respectively with each in several node clusters
Data transfer average speed of the node cluster within D × F period, D is default collection number of days and every day is according to default
The division rule progress period averagely divides, and F is any one day progress period averagely to be divided and sequentially in time to drawing
The period got be numbered after numbering maximum；
S200 is by any one zone user number with each node cluster within the preset numbers period in each day respectively
According to transmission average speed as a set, each set includes a zone user with a node cluster in the pre- of each day
If the data transfer average speed in the numbering period, and travel through Zone Full user；
S300 carries out Polynomial curve-fit to each set obtained after traversal Zone Full user and calculated each
It is expected corresponding to individual set；
S400 will it is expected as one it is expected to gather corresponding to identical zone user, and in each expectation set really
Determine node cluster corresponding to zone user corresponding to greatest hope and greatest hope and greatest hope, each greatest hope is corresponding
Zone user, node cluster corresponding to each greatest hope and preset numbers period correspond and be recorded in master server table
In；
S500 is carried out when receiving when the request instruction that user sends is responded according to request instruction in master server table
Search and matching, when finding and target area user that user respond matches and the request time of request instruction are being preset
When in the numbering period, the node server into node cluster corresponding to the user of target area sends assignment instructions, so that node
Server is treated response user according to assignment instructions selection interdependent node and responded.
Specifically, in the step s 100, obtain each zone user in several zone users respectively with several
Data transfer average speed (i.e. network average speed) of each node cluster within D × F period in node cluster, wherein
D is required default collection number of days when gathered data transmits average speed, and every day according to default division rule (such as
Temporally integral point division or the time least bit division etc.) carry out the period averagely divide, to any one day carry out the period be averaged
After division, the period obtained sequentially in time to division is numbered, and F is the numbering maximum in all numbering, therefore
The time span of each period is TD=24/F (hour).
Sample structure is carried out in step s 200, by any one zone user respectively with each node cluster in each day
Data transfer average speed in the preset numbers period as a set, each set include a zone user with
Data transfer average speed of one node cluster within the preset numbers period in each day, travels through whole zone users, each
Individual zone user corresponds to the quantity identical set of number and node cluster.
In step S300, Polynomial curve-fit is carried out simultaneously to each set obtained after traversal Zone Full user
Calculate and it is expected corresponding to each set.The step is to traveling through all members in each set obtained after Zone Full user
Element carries out Polynomial curve-fit, and calculates each set using the polynomial function obtained after Polynomial curve-fit and correspond to
Expectation.As a kind of specific embodiment, each set obtained after to traversal Zone Full user carries out multinomial
During formula Function Fitting, multinomial can be carried out to each set obtained after traversal Zone Full user according to least square method
Function Fitting, so as to be advantageously implemented to desired quick, accurate calculating corresponding to each set.
In step S400, it will it is expected it is expected set as one corresponding to identical zone user, and in each phase
Hope in set and determine node cluster corresponding to zone user corresponding to greatest hope and greatest hope and greatest hope, by it is each most
It is big it is expected corresponding to zone user, node cluster corresponding to each greatest hope and preset numbers period correspond and be recorded in master
In server table.
Finally, in step S500, when receiving when the request instruction that response user sends, according to request instruction in master
Searched and matched in server table, when finding and target area user that user to be responded matches and request instruction
When request time is within the preset numbers period, the node server into node cluster corresponding to the user of target area sends task
Instruction, responded so that node server selects interdependent node to treat response user according to assignment instructions.According to request instruction
When being searched and matched in master server table, the mesh to match with User IP can be searched according to the User IP of user to be responded
Mark zone user.
To overcome user in distributed virtual environment to access caused by randomness to data interaction between system and user
Influence, the present embodiment proposes above-mentioned distributed file system node selecting method, distributed file system node choosing
Selection method by dividing the desired calculating of data transfer average speed between period and node cluster and zone user to the time,
The currently optimal node cluster of data transmission quality between user to be responded is can determine, so as to realize the node in the node cluster
Server selects a node under it to respond the request instruction of user to be responded, and efficient data are provided for user to be responded
Service.The distributed file system node selecting method that the present embodiment is proposed causes distributed file system by traditional node
Random selection mode is changed into the dynamic node selection mode changed with the change of network environment, improves distributed field system
Unite and the data transmission efficiency between user, ensure that the data interaction quality between system and user, further increase point
The service quality of cloth file system.
As a kind of specific embodiment, distributed file system node selecting method proposed by the invention not only may be used
Using realize be directed to selected under the preset numbers period and node cluster that user data transmission quality to be responded is optimal under node as
User to be responded is serviced, and in each period, the distributed file system node selecting method can be to wait to ring
Serviced using the node under the optimal node cluster of family selection data transmission quality for user to be responded, in above-described embodiment
On basis, include procedure below in step S300 to step S500：
After traveling through Zone Full user, preset numbers period traversal is divided to obtained All Time section, obtained each
Set under period；
Polynomial curve-fit is carried out simultaneously to each set obtained after traversal Zone Full user and All Time section
Calculate and it is expected corresponding to each set；
It will it is expected it is expected set as one corresponding to identical zone user under same time period, and in each expectation
Node cluster corresponding to zone user corresponding to greatest hope and greatest hope and greatest hope is determined in set, by same time
Zone user corresponding to each greatest hope, node cluster corresponding to each greatest hope and corresponding same time period be one by one under section
Corresponding record is in master server table；
When receiving when the request instruction that user sends is responded, searched according to request instruction in master server table
And matching, when finding and target area user that user to be responded matches and match with the request time of request instruction
During target time section, the node server into node cluster corresponding to the user of target area sends assignment instructions, so that node takes
Business device is treated response user according to assignment instructions selection interdependent node and responded.
Present embodiment solves distributed file system data transmission quality between system and user under dynamic network
The problem of unstable, the average speed of data transfer between the division of passage time section and each period lower node cluster and zone user
The desired calculating of degree, the request instruction that can be sent according to user to be responded determine target area user and the target to match
Period, data are carried out so as to treat response user by the node server in node cluster corresponding to selection target zone user
Response, because node cluster corresponding to the user of target area is that data transmission quality is most between user to be responded in current slot
Good node cluster, therefore the node server in the node cluster as corresponding to the user of target area provides high efficiency for user to be responded
Data, services, improve the stability of data transmission quality between user and distributed file system so that distributed document
System randomly chooses mode by traditional node and is changed into the dynamic section changed with the change of the network environment of each period
Point selection mode, the data transmission efficiency between distributed file system and user is improved, ensure that between system and user
Data interaction quality, further increase the service quality of distributed file system.
The node selecting method of the distributed file system of the present invention is described in detail below in conjunction with example：S
Assuming that A is the set of zone user, a1, a2..., anregionFor A element, wherein nregion is zoning
Sum；
Assuming that S is the set of node cluster, s1, s2..., snclusterFor S element, wherein ncluster is the total of node cluster
Number；
D is default collection number of days, and unit is day, is averagely drawn according to the default division rule progress period every day
Point, the period obtained sequentially in time to division is numbered, for example, one day is equally divided into F period, according to the time
Numbering is 1,2 to order respectively ..., and F, the time span of each period is TD=24/F (hour)；
Function mu (m, n, a, s) represent the m days n-th of period (i.e. the numbering of period is n) zone user a to save
Point cluster s data transfer average speed, wherein a ∈ A, s ∈ S, 0 ＜ m≤D, 0 ＜ n≤F；
Set A=μ (m, n, a, s) | and 0 ＜ m≤D } A (n, a, s) is designated as, represent n-th of period of every day (immediately
Between section numbering be n) under zone user a to node cluster s data transfer average speed sample；
Set B=E (A) | and A (n, a, x) ∩ x ∈ S } it is designated as B (n, a) desired collection of the expression with A (n, a, s) for sample
Close；
SetFor zone user aiIp address sets.
According to data with existing sample set A (n, a, s), element μ (m, n, a, s)=(x in A (n, a, s)i, yi), i ∈
(1,2 ..., D), Φ are the function class that multinomial of all numbers no more than l (l≤D) is formed, and now seek polynomial fit functionPolynomial curve-fit is carried out according to least square method so that
Formula (1) is on a0,a1,…alThe function of many variables, therefore above mentioned problem be seek I=I (a0,a1,…al)
Extreme-value problem.Extremum conditions is sought by the function of many variables, can be obtained
I.e.
Formula (3) is on a0,a1,…alSystem of linear equations, be expressed in matrix as
Formula (4) is referred to as normal equation group or normal equation system.
The coefficient matrix of equation group (4) is a symmetric positive definite matrix, therefore existence and unique solution.A is solved from formula (4)k
(k=0,1 ..., l), so as to obtain polynomial fit function
P in formula (5)l(x) formula (1), i.e. p are metl(x) it is required polynomial fit function.By pl(x) and it is expected
Definition calculate sample A (n, a, s) expectation, shown in calculation formula such as formula (6)：
Shift process onto more than, to traversal Zone Full user and All Time section after obtain each gather into
Row Polynomial curve-fit, corresponding polynomial fit function can be solved, can according to the polynomial fit function solved
Corresponding expectation is closed to calculate each.
Next, simulating, verifying has been carried out to the distributed file system node selection algorithm proposed according to the present invention,
The simulating, verifying mainly demonstrates following three aspects：
(1) it whether there is fitting function according to sample；
(2) whether the curve of polynomial fit function is smooth；
(3) the desired existence of fitting function.
Specifically, the sample a in table 1, sample b, sample c, sample d and sample e are in the same area user obtained
The data transfer average speed of user's node cluster different from 5 in the pth period.
Table 1
First, the sample in table 1 is fitted using polynomial fit function, due to computation complexity and calculates run time
Limitation, have selected 3 order polynomials as polynomial fit function, polynomial fit function such as formula (7) institute in this emulation
Show：
F (x)=a1x3+a2x2+a3x+a4 (7)
After table 1 and formula (7) substitution formula (1) (2) (3) (4) are fitted, multinomial corresponding to each sample is obtained
Fitting function：
Sample a：F (x)=0.0008955x3-0.03677x2+0.3445x+99.91 (8)
Sample b：F (x)=- 5.16e-5x3+0.002354x2-0.01298x+95.95 (9)
Sample c：F (x)=7.677e-5x3-0.005957x2+0.1239x+93.63 (10)
Sample d：F (x)=0.0002645x3-0.01583x2+0.2705x+91.92 (11)
Sample e：F (x)=8.575e-5x3-0.004268x2+0.05708x+92.39 (12)
Fig. 4 show the sample a drawn according to formula (8) to formula (12), sample b, sample c, sample d and sample e's
Curve matching figure, Fig. 4 abscissa are track record number, and ordinate is sample value, and the plus sige and fine line in Fig. 4 are respectively
Sample a image and fitted figure picture, empty circles and broken string are sample b image and fitted figure picture, and asterisk and dotted line are sample c
Image and fitted figure picture, the image and fitted figure picture that multiplication sign and point solid line are sample d, solid dot and heavy line are sample e's
Image and fitted figure picture.It is not difficult to find out that sample a, sample b, sample c, sample d and sample e matched curve are smooth from Fig. 4
And sample a matched curve is in highest position in 5 samples.
Formula (8) to formula (12) is substituted into formula (6) respectively, be calculated sample a, sample b, sample c, sample d and
Sample e is corresponding respectively it is expected, as shown in table 2, maximum due to it is expected corresponding to sample a, therefore as the node corresponding to sample a
Cluster selects interdependent node to provide data, services for the zone user within the pth period.
Table 2
| Sample | Sample a | Sample b | Sample c | Sample d | Sample e |
| It is expected | 100.1107 | 96.1207 | 94.2310 | 93.0367 | 92.5520 |
Above simulation result shows that distributed file system node selecting method proposed by the invention can be by multinomial
Formula Function Fitting finds optimum polynomial, and ensure that the flatness of functional image, calculates and it is expected and filter out expectation maximum,
Corresponding node cluster is selected according to expectation maximum, so as to be provided most in different time sections selection for each zone user
The node of good service, distributed file system is randomly choosed mode by traditional node and be changed into the dynamic of change of network environment
State selects node.
Correspondingly, in another embodiment of the present invention, it is proposed that one kind is used for distributed file system node and selected
Master server, for distributed file system node selection master server be configured perform procedure below：
Master server obtain each zone user in several zone users respectively with it is every in several node clusters
Data transfer average speed of one node cluster within D × F period, D is default collection number of days and every day is according to pre-
If division rule carry out the period and averagely divide, F is to carry out the period to any one day averagely to divide and sequentially in time
Numbering maximum after division obtained period is numbered；
Master server is by any one zone user respectively with each node cluster within the preset numbers period in each day
Data transfer average speed as a set, each set includes a zone user and a node cluster in each day
The preset numbers period in data transfer average speed, and travel through Zone Full user；
Master server carries out Polynomial curve-fit to each set obtained after traversal Zone Full user and calculated
It is expected corresponding to each set；
Master server will it is expected it is expected set as one corresponding to identical zone user, and gather in each expectation
Node cluster corresponding to zone user corresponding to middle determination greatest hope and greatest hope and greatest hope, by each greatest hope
Node cluster corresponding to corresponding zone user, each greatest hope and preset numbers period, which correspond, is recorded in master server
In table；
When master server is received when the request instruction that user sends is responded, master server is according to request instruction in main clothes
Searched and matched in business device table, as the target area user and request that master server is found and user to be responded matches
When the request time of instruction is within the preset numbers period, the node in node cluster corresponding to master server to target area user
Server sends assignment instructions, is rung so that node server selects interdependent node to treat response user according to assignment instructions
Should.
Specifically, if in the present embodiment master server obtain each zone user in several zone users respectively with
Data transfer average speed (i.e. network average speed) of each node cluster within D × F period in dry node cluster,
Wherein D is that gathered data transmits default collection number of days required during average speed, and every day advises according to default division
Then (such as temporally integral point division or the division of the time least bit) the progress period is averagely divided, and the period was carried out to any one day
After average division, the period obtained sequentially in time to division is numbered, and F is the numbering maximum in all numbering,
Therefore the time span of each period is TD=24/F (hour).
Master server carries out sample structure, by any one zone user respectively with each node cluster in the default of each day
Data transfer average speed in the numbering period includes a zone user and one as a set, each set
Data transfer average speed of the node cluster within the preset numbers period in each day, travels through whole zone users, each area
The domain quantity identical set for corresponding to number and node cluster per family.
Master server carries out Polynomial curve-fit to each set obtained after traversal Zone Full user and calculated
It is expected corresponding to each set, master server is to traveling through the whole in each set obtained after Zone Full user here
Element carries out Polynomial curve-fit, and calculates each set pair using the polynomial function obtained after Polynomial curve-fit
The expectation answered.As a kind of specific embodiment, each set obtained after to traversal Zone Full user carries out more
During item formula Function Fitting, master server can be according to each set of least square method to being obtained after traversal Zone Full user
Polynomial curve-fit is carried out, so as to be advantageously implemented to desired quick, accurate calculating corresponding to each set.
Master server will it is expected it is expected set as one corresponding to identical zone user, and gather in each expectation
Node cluster corresponding to zone user corresponding to middle determination greatest hope and greatest hope and greatest hope, by each greatest hope
Node cluster corresponding to corresponding zone user, each greatest hope and preset numbers period, which correspond, is recorded in master server
In table.
Finally, when master server is received when the request instruction that user sends is responded, master server is according to request instruction
Searched and matched in master server table, as the target area user that master server is found and user to be responded matches
And the request time of request instruction within the preset numbers period when, in node cluster corresponding to master server to target area user
Node server send assignment instructions so that node server according to assignment instructions select interdependent node treat response user enter
Row response., can be according to user's to be responded when master server is searched and matched in master server table according to request instruction
User IP searches the target area user to match with User IP.
To overcome user in distributed virtual environment to access caused by randomness to data interaction between system and user
Influence, the present embodiment proposes the above-mentioned master server for being used for the selection of distributed file system node, and this is used for distribution
The master server of file system nodes selection to the time by dividing data transfer between period and node cluster and zone user
The desired calculating of average speed, the currently optimal node cluster of data transmission quality between user to be responded is can determine, from
And the request instruction that the node server in the node cluster selects a node under it to respond user to be responded is realized, to wait to ring
Efficient data, services are provided using family.The main service for being used for the selection of distributed file system node that the present embodiment is proposed
Device causes distributed file system to randomly choose mode by traditional node and be changed into the change of network environment and change dynamic
State node selection mode, improves the data transmission efficiency between distributed file system and user, ensure that system and user
Between data interaction quality, further increase the service quality of distributed file system.
As a kind of specific embodiment, the main clothes for being used for the selection of distributed file system node proposed by the invention
Business device can not only realize the node cluster for being selected under the preset numbers period and user data transmission quality to be responded is optimal
Under node serviced for user respond, and in each period, this is used for what distributed file system node selected
Master server can be that user-selected number to be responded enters according to the node under the optimal node cluster of transmission quality for user to be responded
Row service, on the basis of above-described embodiment, master server is configured to perform procedure below：
After master server traversal Zone Full user, preset numbers period traversal is divided to obtained All Time section,
Obtain the set under each period；
Master server carries out multinomial letter to each set obtained after traversal Zone Full user and All Time section
Number intends joint account, and each gathers corresponding expectation；
Master server will it is expected as one it is expected to gather corresponding to identical zone user under same time period, and every
It is expected for one to determine node cluster corresponding to zone user corresponding to greatest hope and greatest hope and greatest hope in set, will
Zone user corresponding to each greatest hope under same time period, node cluster corresponding to each greatest hope and it is corresponding identical when
Between section correspond be recorded in master server table；
When master server is received when the request instruction that user sends is responded, master server is according to request instruction in main clothes
Searched and matched in business device table, as the target area user that master server is found and user to be responded matches and with asking
During the target time section for asking the request time of instruction to match, the section in node cluster corresponding to master server to target area user
Point server sends assignment instructions, is rung so that node server selects interdependent node to treat response user according to assignment instructions
Should.
Present embodiment solves distributed file system data transmission quality between system and user under dynamic network
The problem of unstable, the average speed of data transfer between the division of passage time section and each period lower node cluster and zone user
The desired calculating of degree, the request instruction that can be sent according to user to be responded determine target area user and the target to match
Period, data are carried out so as to treat response user by the node server in node cluster corresponding to selection target zone user
Response, because node cluster corresponding to the user of target area is that data transmission quality is most between user to be responded in current slot
Good node cluster, therefore the node server in the node cluster as corresponding to the user of target area provides high efficiency for user to be responded
Data, services, improve the stability of data transmission quality between user and distributed file system so that distributed document
System randomly chooses mode by traditional node and is changed into the dynamic section changed with the change of the network environment of each period
Point selection mode, the data transmission efficiency between distributed file system and user is improved, ensure that between system and user
Data interaction quality, further increase the service quality of distributed file system.
It is used for the implementation method of the concrete function of the master server of distributed file system node selection in the present invention, can be with
With reference to the implementation method described in above-mentioned distributed file system node selecting method embodiment, here is omitted.
Each technical characteristic of embodiment described above can be combined arbitrarily, to make description succinct, not to above-mentioned reality
Apply all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, the scope that this specification is recorded all is considered to be.
Embodiment described above only expresses the several embodiments of the present invention, and its description is more specific and detailed, but simultaneously
Can not therefore it be construed as limiting the scope of the patent.It should be pointed out that come for one of ordinary skill in the art
Say, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (6)
1. a kind of distributed file system node selecting method, it is characterised in that comprise the following steps：
Obtain each described zone user in several zone users respectively with described in each in several node clusters
Data transfer average speed of the node cluster within D × F period, the D is default collection number of days and every day is according to pre-
If division rule carry out the period averagely divide, the F be to any one day carry out the period averagely divide and according to the time
Order division obtained period is numbered after numbering maximum；
By any one of zone user number with node cluster each described within the preset numbers period in each day respectively
Include a zone user and a node as a set, each described set according to transmission average speed
The data transfer average speed of the cluster within the preset numbers period in each day, and travel through all regions and use
Family；
Polynomial curve-fit is carried out to each the described set obtained after the traversal all zone user and calculated each
It is expected corresponding to the individual set；
It will it is expected it is expected set as one corresponding to zone user described in identical, and in each described expectation set really
Determine node cluster corresponding to zone user corresponding to greatest hope and the greatest hope and the greatest hope, will be each described
A pair of zone user corresponding to greatest hope, node cluster corresponding to each greatest hope and preset numbers period 1
It should be recorded in master server table；
When receiving when the request instruction that user sends is responded, carried out according to the request instruction in the master server table
Search and match, when finding the request with the target area user to match when response user and the request instruction
Between within the preset numbers period when, node server into node cluster corresponding to the target area user, which is sent, appoints
Business instruction, so that the node server selects interdependent node to ring the user to be responded according to the assignment instructions
Should.
2. distributed file system node selecting method according to claim 1, it is characterised in that described pair of traversal is all
Each the described set obtained after the zone user carries out Polynomial curve-fit and calculates each described set correspondingly
Expectation the step of to it is described when receive wait respond user send request instruction when, according to the request instruction in the master
The step of being searched and matched in server table includes procedure below：
Traversal all after the zone user, preset numbers period traversal is divided the obtained whole period,
Obtain the set under each period；
To traversal, all the zone user and each the described set all obtained after the period carry out multinomial letter
It is expected corresponding to each described set of number plan joint account；
It will it is expected corresponding to zone user described in identical under same time period as an expectation set, and at each
It is described it is expected to determine zone user and the greatest hope pair corresponding to the greatest hope and the greatest hope in set
The node cluster answered, zone user, each greatest hope corresponding to each greatest hope under same time period is corresponding
Node cluster and the corresponding same time period correspond and be recorded in the master server table；
When receiving when the request instruction that user sends is responded, carried out according to the request instruction in the master server table
Search and matching, when finding the target area user to match with the user to be responded and the request with the request instruction
During the target time section that the time matches, the node server into node cluster corresponding to the target area user sends task
Instruction, so that the node server selects interdependent node to respond the user to be responded according to the assignment instructions.
3. distributed file system node selecting method according to claim 1 or 2, it is characterised in that
Polynomial function is carried out to each the described set obtained after the traversal all zone user according to least square method
Fitting.
4. a kind of master server for the selection of distributed file system node, it is characterised in that master server is configured to perform
Procedure below：
The master server obtain each described zone user in several zone users respectively with several node clusters
Data transfer average speed of each described node cluster within D × F period, the D is default collection number of days and every
Averagely divided according to the default division rule progress period within one day, the F is that any one day progress period was averagely drawn
Point and sequentially in time to divide obtained period be numbered after numbering maximum；
The master server by any one of zone user respectively with preset numbers of the node cluster each described in each day
Data transfer average speed in period as a set, each described set include a zone user with
The data transfer average speed of one node cluster within the preset numbers period in each day, and travel through whole
The zone user；
The master server carries out polynomial function plan to each the described set obtained after the traversal all zone user
It is expected corresponding to each described set of joint account；
The master server will it is expected as one it is expected to gather corresponding to zone user described in identical, and described in each
It is expected to determine node corresponding to zone user corresponding to greatest hope and the greatest hope and the greatest hope in set
Cluster, by node cluster and the default volume corresponding to zone user corresponding to each greatest hope, each greatest hope
Number period corresponds and is recorded in master server table；
When the master server is received when the request instruction that user sends is responded, the master server refers to according to the request
Order is searched and matched in the master server table, is matched when the master server is found with the user to be responded
Target area user and the request instruction request time within the preset numbers period when, the master server to
Node server in node cluster corresponding to the target area user sends assignment instructions so that the node server according to
The assignment instructions selection interdependent node responds to the user to be responded.
5. the master server according to claim 4 for the selection of distributed file system node, it is characterised in that
The master server traversal all after the zone user, preset numbers period traversal is divided obtained whole
The period, obtain the set under each period；
The master server is to the traversal all zone user and each described set all obtained after the period
Carry out Polynomial curve-fit and calculate it is expected corresponding to each described set；
The master server will it is expected to collect as an expectation corresponding to zone user described in identical under same time period
Close, and zone user and institute corresponding to the greatest hope and the greatest hope are determined in each described expectation set
Node cluster corresponding to greatest hope is stated, by zone user corresponding to each greatest hope under same time period, each described
Node cluster corresponding to greatest hope and the corresponding same time period are corresponded and are recorded in the master server table；
When the master server is received when the request instruction that user sends is responded, the master server refers to according to the request
Order is searched and matched in the master server table, is matched when the master server is found with the user to be responded
Target area user and match with the request time of the request instruction target time section when, the master server is to institute
The node server stated in node cluster corresponding to the user of target area sends assignment instructions, so that the node server is according to institute
Assignment instructions selection interdependent node is stated to respond the user to be responded.
6. the master server for being used for the selection of distributed file system node according to claim 4 or 5, it is characterised in that
The master server according to least square method to obtained after the traversal all zone users each it is described gather into
Row Polynomial curve-fit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710821141.3A CN107508906B (en) | 2017-09-13 | 2017-09-13 | Distributed file system node selection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710821141.3A CN107508906B (en) | 2017-09-13 | 2017-09-13 | Distributed file system node selection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107508906A true CN107508906A (en) | 2017-12-22 |
| CN107508906B CN107508906B (en) | 2020-03-13 |
Family
ID=60696077
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710821141.3A Active CN107508906B (en) | 2017-09-13 | 2017-09-13 | Distributed file system node selection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107508906B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101159745A (en) * | 2007-11-08 | 2008-04-09 | 中国传媒大学 | Intelligent node elastic overlapping network for P2P application with copyright management function |
| CN101252495A (en) * | 2008-03-06 | 2008-08-27 | 北京航空航天大学 | Wireless sensor network local type node managing method |
| CN103152187A (en) * | 2011-12-06 | 2013-06-12 | 中国科学院声学研究所 | Aggregation method and system of nodes mapping and fusing much information |
| US20140047084A1 (en) * | 2012-08-07 | 2014-02-13 | Advanced Micro Devices, Inc. | System and method for modifying a hardware configuration of a cloud computing system |
-
2017
- 2017-09-13 CN CN201710821141.3A patent/CN107508906B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101159745A (en) * | 2007-11-08 | 2008-04-09 | 中国传媒大学 | Intelligent node elastic overlapping network for P2P application with copyright management function |
| CN101252495A (en) * | 2008-03-06 | 2008-08-27 | 北京航空航天大学 | Wireless sensor network local type node managing method |
| CN103152187A (en) * | 2011-12-06 | 2013-06-12 | 中国科学院声学研究所 | Aggregation method and system of nodes mapping and fusing much information |
| US20140047084A1 (en) * | 2012-08-07 | 2014-02-13 | Advanced Micro Devices, Inc. | System and method for modifying a hardware configuration of a cloud computing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107508906B (en) | 2020-03-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tong et al. | Online minimum matching in real-time spatial data: experiments and analysis | |
| CN107241319B (en) | Distributed network crawler system based on VPN and scheduling method | |
| Wang et al. | Service placement for collaborative edge applications | |
| CN106708625A (en) | Minimum-cost maximum-flow based large-scale resource scheduling system and minimum-cost maximum-flow based large-scale resource scheduling method | |
| CN110519090B (en) | Method and system for allocating accelerator cards of FPGA cloud platform and related components | |
| Tang et al. | Coordinate-based efficient indexing mechanism for intelligent IoT systems in heterogeneous edge computing | |
| Khanli et al. | FRDT: footprint resource discovery tree for grids | |
| CN111274485A (en) | Personalized recommendation method based on community discovery | |
| Malik et al. | Latency based group discovery algorithm for network aware cloud scheduling | |
| US20160342899A1 (en) | Collaborative filtering in directed graph | |
| CN113228059A (en) | Cross-network-oriented representation learning algorithm | |
| Wu et al. | Optimizing end-to-end performance of data-intensive computing pipelines in heterogeneous network environments | |
| CN112702390B (en) | Block chain-based networking method and device for intelligent contract resources | |
| Toce et al. | An efficient hypercube labeling schema for dynamic peer-to-peer networks | |
| CN107508906A (en) | Distributed file system node selecting method and master server | |
| Xu et al. | DLCD-CCE: A local community detection algorithm for complex IoT networks | |
| CN101442466B (en) | Superpose network and implementing method | |
| Phoha et al. | Faster Web page allocation with neural networks | |
| Toda et al. | Autonomous and distributed construction of locality aware skip graph | |
| Alzahrani et al. | Energy-aware virtual network embedding approach for distributed cloud | |
| Ali et al. | Using the analytical network process for controller placement in software defined networks (poster) | |
| Barzegar et al. | A new method on resource management in grid computing systems based on QoS and semantics | |
| Lu et al. | High-revenue Online Provisioning for Virtual Clusters in Multi-tenant Cloud Data Center Network | |
| Gilles et al. | Simulating social network formation: A case-based decision theoretic model | |
| Cui et al. | Topological embedding feature based resource allocation in network virtualization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |