CN106411605B - A kind of meshed network self-organizing method, device, server and system - Google Patents
A kind of meshed network self-organizing method, device, server and system Download PDFInfo
- Publication number
- CN106411605B CN106411605B CN201610963725.XA CN201610963725A CN106411605B CN 106411605 B CN106411605 B CN 106411605B CN 201610963725 A CN201610963725 A CN 201610963725A CN 106411605 B CN106411605 B CN 106411605B
- Authority
- CN
- China
- Prior art keywords
- node
- meshed network
- task
- port numbers
- sent
- 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
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The invention discloses a kind of meshed network self-organizing method, device, server and systems.Wherein, method includes: the task start instruction for receiving main controlled node and sending；Choose the port numbers for being used for the task；This is returned into the main controlled node from the host name of node and the port numbers of selection, so that the main controlled node generates meshed network figure according to respectively starting from node for task and the host name and port numbers of return；Receive the meshed network figure that the main controlled node is sent.The distribution of port numbers is changed to respectively by main controlled node from node by the technical solution, the port numbers for avoiding main controlled node distribution are clashed with from the used port numbers of node, the meshed network figure generated simultaneously can be used for the management of meshed network and establish from the connection between node, both use demand is met, while improving the success rate of meshed network building.
Description
Technical field
The present invention relates to technical field of the computer network, and in particular to a kind of meshed network self-organizing method, device, service
Device and system.
Background technique
There are multiple nodes in distributed type assemblies, can be run on these nodes and execute same task parallel.In many situations
Under, executing tasks parallelly also needs to be communicated between node and node, therefore the tissue of meshed network is very crucial asks
Topic.In the prior art, this when establishing, is usually executed in task for the slave node distribution of each execution task by main controlled node
Port numbers used in the process of business respectively can know that other execute the process of the task from node and are made from node in this way
Port numbers, thus and other from node establish connection.But the port numbers of distribution are actually and respectively from one a pair of node
It answers, if main controlled node does not know used port numbers on a certain node, the node is assigned in assignment of port numbers
The port numbers used just will affect the starting of task.For example, the task A run on node 1 uses port 8080, in master
Control node is that node 1 specifies port 8080 when distributing port numbers used in task B again, then just causing port collision.
Summary of the invention
In view of the above problems, it proposes on the present invention overcomes the above problem or at least be partially solved in order to provide one kind
State meshed network self-organizing method, device, server and the system of problem.
According to one aspect of the present invention, a kind of meshed network self-organizing method is provided, comprising:
Receive the task start instruction that main controlled node is sent；
Choose the port numbers for being used for the task；
This is returned into the main controlled node from the host name of node and the port numbers of selection, so that the main controlled node root
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated；
Receive the meshed network figure that the main controlled node is sent.
Optionally, described choose includes: for the port numbers of the task
From this from node currently unappropriated port numbers, a port number is randomly selected.
Optionally, the meshed network figure for receiving the main controlled node transmission includes:
The request for obtaining meshed network figure is periodically sent to the main controlled node, receives the main controlled node according to the request
The meshed network figure of return；
And/or
Receive the meshed network figure that the main controlled node actively issues.
Optionally, this method further include:
According to the meshed network figure that the main controlled node is sent, in the meshed network figure other are one or more from section
Point establishes connection.
Optionally, the task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask
It include: the subtask parameter server and/or the subtask worker.
Another aspect according to the present invention provides a kind of meshed network self-organizing method, comprising:
According to the mission bit stream of input, task start instruction is sent to one or more from node；
Receive the host name and port numbers respectively returned from node；
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated；
The meshed network figure is sent to one or more from node.
Optionally, described the meshed network figure is sent to one or more from node to include:
When receive from node send acquisition meshed network figure request when, by the meshed network figure be sent to this from
Node；
And/or
The meshed network figure is sent to connect with this main controlled node it is all from node.
Optionally, the mission bit stream is the mission bit stream of deep learning task；The mission bit stream includes: for executing
The number of nodes of deep learning task, deep learning subtask type, all types of subtask quantity.
Optionally, the mission bit stream according to input, sending task start instruction from node to one or more includes:
From all number of nodes for being selected from node and being used to execute deep learning task being connect with this main controlled node
It is comparable from node；
According to deep learning subtask type and all types of subtask quantity, determines and appoint what is respectively started from node
Business；
To the slave node of each selection send in starting from node for the task corresponding task start instruction.
Another aspect according to the present invention provides a kind of meshed network self-organizing device, wherein the device is deployed in point
On the slave node of cloth cluster, comprising:
Communication unit, the task start instruction sent suitable for receiving node self-organization of network server；
Port selection unit, suitable for choosing the port numbers for being used for the task；
The communication unit is further adapted for return to where the present apparatus from the host name of node and the port numbers of selection described
Meshed network hoc service device, so that the meshed network hoc service device is according to respectively starting from node for task and each section
The host name and port numbers that spot net self-organizing device returns generate meshed network figure；And it is suitable for receiving the meshed network
The meshed network figure that hoc service device is sent.
Optionally, the port selection unit, suitable for the current unappropriated port numbers where the present apparatus from node
In, randomly select a port number.
Optionally, the communication unit is suitable for periodically sending acquisition node net to the meshed network hoc service device
The request of network figure receives the meshed network figure that the meshed network hoc service device is returned according to the request；And/or it receives
The meshed network figure that the meshed network hoc service device actively issues.
Optionally, the communication unit is further adapted for the meshed network sent according to the meshed network hoc service device
Figure establishes connection from the meshed network self-organizing device on node with other one or more in the meshed network figure.
Optionally, the task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask
It include: the subtask parameter server and/or the subtask worker.
According to the present invention in another aspect, providing a kind of meshed network hoc service device, wherein the server disposition
On the main controlled node of distributed type assemblies, comprising:
Communication unit, suitable for the mission bit stream according to input, to one or more from node on meshed network self-organizing
Device sends task start instruction；Receive host name and port numbers that each meshed network self-organizing device returns；
Meshed network figure generation unit, suitable for being returned according to the task and each meshed network self-organizing device that respectively start from node
The host name and port numbers returned generate meshed network figure；
The communication unit is further adapted for for the meshed network figure being sent to one or more from the meshed network on node
Self-organizing device.
Optionally, the communication unit, suitable for receiving obtaining from the meshed network self-organizing device transmission on node
When taking the request of meshed network figure, the meshed network figure is sent to this from the meshed network self-organizing device on node；With/
Or, the meshed network figure to be sent to all meshed networks from node connecting with the main controlled node where book server
Self-organizing device
Optionally, the mission bit stream is the mission bit stream of deep learning task；The mission bit stream includes: for executing
The number of nodes of deep learning task, deep learning subtask type, all types of subtask quantity.
Optionally, the server further include:
Scheduling unit, suitable for from connect with the main controlled node where book server it is all from node selection be used for hold
The number of nodes of row deep learning task is comparable from node；According to deep learning subtask type and all types of subtask numbers
Amount is determined in respectively starting from node for task；
The communication unit, suitable for the meshed network self-organizing device on the slave node to selection send with this from node
The corresponding task start instruction of the task of upper starting.
According to the present invention in another aspect, providing a kind of meshed network self-organizing system, wherein the system include one
Or multiple meshed network self-organizing devices as described in any one of the above embodiments and meshed network self-organizing as described in any one of the above embodiments
Server.
It can be seen from the above, technical solution of the present invention, from node in the task start instruction for receiving main controlled node transmission
Afterwards, the port numbers for being used for the task are actively chosen, this is returned into main controlled node from the host name of node and the port numbers of selection,
So that main controlled node generates meshed network figure according to respectively starting from node for task and the host name and port numbers of return.The skill
Art scheme the distribution of port numbers is changed to respectively avoid from node by main controlled node the port numbers of main controlled node distribution with from section
The used port numbers of point clash, while the meshed network figure generated can be used for the management of meshed network and between node
Connection establish, both meet use demand, at the same improve meshed network building success rate.
The above description is only an overview of the technical scheme of the present invention, in order to better understand the technical means of the present invention,
And it can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage can
It is clearer and more comprehensible, the followings are specific embodiments of the present invention.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention
Limitation.And throughout the drawings, the same reference numbers will be used to refer to the same parts.In the accompanying drawings:
Fig. 1 shows a kind of flow diagram of meshed network self-organizing method according to an embodiment of the invention；
Fig. 2 shows the flow diagrams of another meshed network self-organizing method according to an embodiment of the invention；
Fig. 3 shows a kind of structural schematic diagram of meshed network self-organizing device according to an embodiment of the invention；
Fig. 4 shows a kind of structural schematic diagram of meshed network hoc service device according to an embodiment of the invention；
Fig. 5 shows a kind of structural schematic diagram of meshed network self-organizing system according to an embodiment of the invention.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
Fig. 1 shows a kind of flow diagram of meshed network self-organizing method according to an embodiment of the invention, such as
Shown in Fig. 1, this method comprises:
Step S110 receives the task start instruction that main controlled node is sent.
Step S120 chooses the port numbers for being used for the task.
This is returned to main controlled node from the host name of node and the port numbers of selection, so that main controlled node by step S130
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated.
Step S140 receives the meshed network figure that main controlled node is sent.
As it can be seen that method shown in FIG. 1 is actively selected from node after receiving the task start instruction that main controlled node is sent
The port numbers in the task are taken, this are returned into main controlled node from the host name of node and the port numbers of selection, so that master control
Node generates meshed network figure according to respectively starting from node for task and the host name and port numbers of return.The technical solution will
The distribution of port numbers is changed to by main controlled node respectively from node, and the port numbers for avoiding main controlled node distribution have been used with from node
Port numbers clash, while generate meshed network figure can be used for the management of meshed network and built from the connection between node
It is vertical, use demand is both met, while improving the success rate of meshed network building.
In one embodiment of the invention, in method shown in FIG. 1, choose for the task port numbers include: from
This randomly selects a port number from node currently unappropriated port numbers.
It is usually 65535 circulations from the port numbers on node, is chosen from unappropriated port numbers at random, it is high-efficient,
Port collision will not be generated simultaneously.
In one embodiment of the invention, in method shown in FIG. 1, the meshed network figure packet that main controlled node is sent is received
It includes: periodically sending the request for obtaining meshed network figure to main controlled node, receive the node net that main controlled node is returned according to the request
Network figure；And/or receive the meshed network figure that main controlled node actively issues.
For example, sending the request for obtaining meshed network figure to main controlled node every 10 minutes, or saved by main controlled node
After spot net figure changes, updated meshed network figure is issued to relevant from node.
Respectively from node obtain meshed network figure a major reason be run from node in many cases, into
Journey needs are communicated with other from the process on node.Therefore in one embodiment of the invention, method shown in FIG. 1 is also
Include: the meshed network figure sent according to main controlled node, is established with other one or more in the meshed network figure from node
Connection.
In one embodiment of the invention, in method shown in FIG. 1, task start instruction is deep learning subtask
Enabled instruction；Deep learning subtask includes: the subtask parameter server and/or the subtask worker.In this instance,
Parameter server needs to receive the parameter of the subtask worker submission being calculated as parameter server.
Fig. 2 shows the flow diagram of another meshed network self-organizing method according to an embodiment of the invention,
As shown in Fig. 2, this method comprises:
Step S210 sends task start instruction to one or more from node according to the mission bit stream of input.
Step S220 receives the host name and port numbers respectively returned from node.
Step S230 generates meshed network figure according to the host name and port numbers of the task and return that respectively start from node.
Meshed network figure is sent to one or more from node by step S240.
In one embodiment of the invention, in method shown in Fig. 2, by meshed network figure be sent to it is one or more from
Node includes: that meshed network figure is sent to this from section when receiving the request of the acquisition meshed network figure sent from node
Point；And/or meshed network figure is sent to connect with this main controlled node it is all from node.
The distribution method of two kinds of meshed network figures is provided in the present embodiment, can be used in combination, but representative pair
Updated meshed network figure can also be only sent to this when meshed network figure is changed by the limitation of distribution method
Update relevant node.For example, task A has increased two execution nodes, node 13 and node 14 newly, task A original executes node and is
Node 1 and node 2, then only needing updated meshed network figure being sent to node 1,3,13 and 14.Certainly, meshed network
Figure can also generate corresponding component according to each task, only need to distribute them to the figure in this way when issuing meshed network figure
In relevant node.
In one embodiment of the invention, in method shown in Fig. 2, mission bit stream is that the task of deep learning task is believed
Breath；Mission bit stream includes: for executing the number of nodes of deep learning task, deep learning subtask type, all types of sons
Task quantity.
Deep learning task is to carry out the submission of calculating task in graph form, these tasks when being executed can be further
Multiple operations are divided into, each operation includes one or more subtasks.Subtask type includes one of following or a variety of:
The subtask parameter server, the subtask worker.
For example, TensorFlow is exactly the deep learning library of a open source.Tensor (tensor) means N-dimensional array,
Flow (stream) means the calculating based on data flow diagram, and TensorFlow flow to the other end from one end of image for tensor and calculates
Process.The deep learning library can be integrated with Spark big data Computational frame, i.e., using a TensorFlow task as
One Spark task is submitted, that is, so-called deep learning task above.Deep learning mission bit stream can also include
It is one of following or a variety of: to execute the calculating figure of deep learning；The deep learning library that executing deep learning task need to call connects
Mouthful；Data address for deep learning task；The preservation address of implementing result data.
In one embodiment of the invention, in the above method, according to the mission bit stream of input, to one or more from section
It includes: from all selections from node connecting with this main controlled node and for executing deep learning that point, which sends task start instruction,
The number of nodes of task is comparable from node；According to deep learning subtask type and all types of subtask quantity, determine
Respectively starting from node for task；It sends to the slave node of each selection and is opened in the corresponding task of starting from node for the task
Dynamic instruction.
By taking a deep learning task as an example, if needing to call depth according to the mission bit stream of the deep learning task
Learning database starts 2 subtasks parameter server and 2 subtasks worker, and this four subtasks are respectively four
It is a to be executed from node, then just first determining the subtask executed in each task, then starting phase is sent from node to each
The instruction for the subtask answered.
It is mentioned above, deep learning library can be integrated with Spark big data Computational frame, i.e. distributed type assemblies can be with
For Spark cluster.Spark cluster can also carry out scheduling, job management and the resource management of task by Yarn.Yarn can be with
Submission of the front end page for task is provided for user, therefore in one embodiment of the invention, the deep learning of submission is appointed
Business can be through front end page input.After task start, the front end page that user can also provide according to Yarn, in real time
The treatment situation for checking task carries out task the operation such as to kill.Since Spark cluster can also carry out task by Yarn
Scheduling, job management and resource management, therefore in above-described embodiment, all being selected from node of Cong Yuben main controlled node connection
With the number of nodes for executing deep learning task is comparable can also be requested from node by sending to Yarn, obtain current
More idle node executes deep learning task.That is: it sends to the node scheduling device of distributed type assemblies for executing the depth
Spend the number of nodes of learning tasks, and the information of multiple nodes of receiving node scheduler return.
The example of the meshed network figure generated for a deep learning task has been illustrated below:
{PS:[node1:8080,node2:8080]worker:[node3:9090,node4:9090]}
This means that the subtask parameter server is started on 8080 ports of node 1, the 8080 of node 2
The subtask parameter server is started on port；The subtask worker is started on 9090 ports of node 3, is being saved
The subtask worker is started on 9090 ports of point 4.Next it needs that meshed network figure is actively handed down to these from node,
Or according to by respectively sending from node from favourite network list acquisition request, meshed network figure is handed down to these nodes.Example
Such as, the subtask worker started on 9090 ports of node 3 can respectively with start on 8080 ports of node 1
The subtask parameter server and the subtask parameter server started on 8080 ports of node 2 are established
Connection.
These can start a Driver process by Spark, while starting one after the submission of deep learning task
Scheduler dispatches process, and building, management and the distribution of meshed network figure are realized by the process.
Specifically, obtaining from the file system of distributed type assemblies for the data of the deep learning task includes: basis
For the data address of deep learning task, the data structure of the deep learning task will be used in the file system of distributed type assemblies
It builds as elasticity distribution formula data set RDD object；The data-pushing for the deep learning task that will acquire is appointed to corresponding son
Carrying out executing in business includes: that RDD object is pushed to each node respectively, by each node by RDD Object Push in the node
On the subtask of starting.
By taking Spark distributed type assemblies as an example, data are stored in HDFS (Hadoop Distributed File
System, Hadoop distributed file system) on.In operation data, it is configured to a RDD accordingly
(resilientdistributed dataset, elasticity distribution formula data set) object.RDD object can be multiplexed, if depth
Data used in learning tasks have been built as RDD object, then natural, there is no need to execute the step.Using these data
When, pushed it to by pipeline (pipe) on the slave node where each task, by each node by RDD Object Push to this from
On the subtask started in node.For deep learning task in the above example includes two subtasks worker, need RDD
A part of object is pushed on node 3, and another part is pushed on node 4, to realize distributed treatment deep learning
Task.
Fig. 3 shows a kind of structural schematic diagram of meshed network self-organizing device according to an embodiment of the invention, should
Device can be deployed on the slave node of distributed type assemblies.As shown in figure 3, meshed network self-organizing device 300 includes:
Communication unit 310, the task start instruction sent suitable for receiving node self-organization of network server.
Port selection unit 320, suitable for choosing the port numbers for being used for the task.
Communication unit 310 is further adapted for that node will be returned to from the host name of node and the port numbers of selection where the present apparatus
Self-organization of network server so that meshed network hoc service device according to respectively from node start task and each meshed network from
The host name and port numbers that tissue device returns generate meshed network figure；And it is suitable for receiving node self-organization of network server
The meshed network figure of transmission.
As it can be seen that device shown in Fig. 3 is receiving what main controlled node was sent from node by the mutual cooperation of each unit
After task start instruction, the port numbers for being used for the task are actively chosen, this is returned from the host name of node and the port numbers of selection
Back to main controlled node, so that main controlled node generates section according to respectively starting from node for task and the host name and port numbers of return
Spot net figure.The distribution of port numbers is changed to respectively avoid main controlled node distribution from node by main controlled node by the technical solution
Port numbers clashed with from the used port numbers of node, while generate meshed network figure can be used for meshed network
It manages and is established from the connection between node, both meet use demand, while improving the success rate of meshed network building.
In one embodiment of the invention, in device shown in Fig. 3, port selection unit 320 is suitable for from present apparatus institute
In the current unappropriated port numbers from node, a port number is randomly selected.
In one embodiment of the invention, in device shown in Fig. 3, communication unit 310 is suitable for periodically to meshed network
Hoc service device sends the request for obtaining meshed network figure, and receiving node self-organization of network server is returned according to the request
Meshed network figure；And/or the meshed network figure that receiving node self-organization of network server actively issues.
In one embodiment of the invention, in device shown in Fig. 3, communication unit 310 is further adapted for according to meshed network
The meshed network figure that hoc service device is sent, in the meshed network figure other are one or more from the node net on node
Network self-organizing device 300 establishes connection.
In one embodiment of the invention, in device shown in Fig. 3, task start instruction is deep learning subtask
Enabled instruction；Deep learning subtask includes: the subtask parameter server and/or the subtask worker.
Fig. 4 shows a kind of structural schematic diagram of meshed network hoc service device according to an embodiment of the invention,
The server can be deployed on the main controlled node of distributed type assemblies.As shown in figure 4, meshed network hoc service device 400 wraps
It includes:
Communication unit 410, suitable for the mission bit stream according to input, to one or more from node on meshed network from group
It knits device and sends task start instruction；Receive host name and port numbers that each meshed network self-organizing device returns；
Meshed network figure generation unit 420, suitable for according to respectively starting from node for task and each meshed network self-organizing dress
The host name and port numbers of return are set, meshed network figure is generated；
Communication unit 410 is further adapted for for meshed network figure being sent to one or more from the meshed network on node from group
Knit device.
In one embodiment of the invention, in server shown in Fig. 4, communication unit 410, suitable for receiving from section
When the request for the acquisition meshed network figure that the meshed network self-organizing device on point is sent, meshed network figure is sent to this from section
Meshed network self-organizing device on point；And/or meshed network figure is sent to and is connect with the main controlled node where book server
All meshed network self-organizing devices from node
In one embodiment of the invention, in server shown in Fig. 4, mission bit stream is the task of deep learning task
Information；Mission bit stream includes: for executing the number of nodes of deep learning task, deep learning subtask type, all types of
Subtask quantity.
In one embodiment of the invention, in server shown in Fig. 4 further include: scheduling unit 430, be suitable for from this
The all of main controlled node connection where server select and the number of nodes phase for executing deep learning task from node
When slave node；According to deep learning subtask type and all types of subtask quantity, determines and respectively to start from node
Task；Communication unit 310 sends suitable for the meshed network self-organizing device on the slave node to selection and is opened from node at this
The corresponding task start instruction of dynamic task.
It should be noted that the specific embodiment of above-mentioned each device and server example and aforementioned corresponding method are implemented
The specific embodiment of example is similar, and details are not described herein.It is slightly different, node respectively can be not only deployed with from node
Self-organization of network device, can be with the executive device of deployment task.Meshed network can be not only deployed on main controlled node from group
Server is knitted, it can be with the control server of deployment task.Certainly, these servers can be integrated each by function and be used as one
A server realizes that same each device from node can also be integrated as a device by function and be realized.
Fig. 5 shows a kind of structural schematic diagram of meshed network self-organizing system according to an embodiment of the invention, such as
Shown in Fig. 5, meshed network self-organizing system 500 includes one or more meshed network self-organizings as in above-mentioned any embodiment
Meshed network hoc service device 400 in device 300 and such as above-mentioned any embodiment.
In conclusion technical solution of the present invention, from node after receiving the task start instruction that main controlled node is sent,
The port numbers for being used for the task are actively chosen, this is returned into main controlled node from the host name of node and the port numbers of selection, with
Make main controlled node according to respectively starting from node for task and the host name and port numbers of return, generates meshed network figure.The technology
Scheme the distribution of port numbers is changed to respectively avoid from node by main controlled node the port numbers of main controlled node distribution with from node
Used port numbers clash, while the meshed network figure generated can be used for the management of meshed network and between node
Connection is established, and use demand is both met, while improving the success rate of meshed network building.
It should be understood that
Algorithm and display be not inherently related to any certain computer, virtual bench or other equipment provided herein.
Various fexible units can also be used together with teachings based herein.As described above, it constructs required by this kind of device
Structure be obvious.In addition, the present invention is also not directed to any particular programming language.It should be understood that can use various
Programming language realizes summary of the invention described herein, and the description done above to language-specific is to disclose this hair
Bright preferred forms.
In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that implementation of the invention
Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this specification.
Similarly, it should be understood that in order to simplify the disclosure and help to understand one or more of the various inventive aspects,
Above in the description of exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes
In example, figure or descriptions thereof.However, the disclosed method should not be interpreted as reflecting the following intention: i.e. required to protect
Shield the present invention claims features more more than feature expressly recited in each claim.More precisely, as following
Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
Thus the claims for following specific embodiment are expressly incorporated in the specific embodiment, wherein each claim itself
All as a separate embodiment of the present invention.
Those skilled in the art will understand that can be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more devices different from this embodiment.It can be the module or list in embodiment
Member or component are combined into a module or unit or component, and furthermore they can be divided into multiple submodule or subelement or
Sub-component.Other than such feature and/or at least some of process or unit exclude each other, it can use any
Combination is to all features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed
All process or units of what method or apparatus are combined.Unless expressly stated otherwise, this specification is (including adjoint power
Benefit require, abstract and attached drawing) disclosed in each feature can carry out generation with an alternative feature that provides the same, equivalent, or similar purpose
It replaces.
In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments
In included certain features rather than other feature, but the combination of the feature of different embodiments mean it is of the invention
Within the scope of and form different embodiments.For example, in the following claims, embodiment claimed is appointed
Meaning one of can in any combination mode come using.
Various component embodiments of the invention can be implemented in hardware, or to run on one or more processors
Software module realize, or be implemented in a combination thereof.It will be understood by those of skill in the art that can be used in practice
Microprocessor or digital signal processor (DSP) realize meshed network self-organizing device according to an embodiment of the present invention, clothes
The some or all functions of business device and some or all components in system.The present invention is also implemented as executing this
In described method some or all device or device programs (for example, computer program and computer program
Product).It is such to realize that program of the invention can store on a computer-readable medium, it either can have one or more
The form of a signal.Such signal can be downloaded from an internet website to obtain, be perhaps provided on the carrier signal or with
Any other form provides.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and ability
Field technique personnel can be designed alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between parentheses should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not
Element or step listed in the claims.Word "a" or "an" located in front of the element does not exclude the presence of multiple such
Element.The present invention can be by means of including the hardware of several different elements and being come by means of properly programmed computer real
It is existing.In the unit claims listing several devices, several in these devices can be through the same hardware branch
To embody.The use of word first, second, and third does not indicate any sequence.These words can be explained and be run after fame
Claim.
Embodiment of the invention discloses A1, a kind of meshed network self-organizing method, wherein this method comprises:
Receive the task start instruction that main controlled node is sent；
Choose the port numbers for being used for the task；
This is returned into the main controlled node from the host name of node and the port numbers of selection, so that the main controlled node root
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated；
Receive the meshed network figure that the main controlled node is sent.
A2, method as described in a1, wherein described choose include: for the port numbers of the task
From this from node currently unappropriated port numbers, a port number is randomly selected.
A3, method as described in a1, wherein described to receive the meshed network figure that the main controlled node is sent and include:
The request for obtaining meshed network figure is periodically sent to the main controlled node, receives the main controlled node according to the request
The meshed network figure of return；
And/or
Receive the meshed network figure that the main controlled node actively issues.
A4, method as described in a1, wherein this method further include:
According to the meshed network figure that the main controlled node is sent, in the meshed network figure other are one or more from section
Point establishes connection.
A5, method as described in a1, wherein
The task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask includes:
The subtask parameter server and/or the subtask worker.
The embodiment of the present invention also discloses B6, a kind of meshed network self-organizing method, wherein this method comprises:
According to the mission bit stream of input, task start instruction is sent to one or more from node；
Receive the host name and port numbers respectively returned from node；
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated；
The meshed network figure is sent to one or more from node.
B7, the method as described in B6, wherein described that the meshed network figure is sent to one or more from node packet
It includes:
When receive from node send acquisition meshed network figure request when, by the meshed network figure be sent to this from
Node；
And/or
The meshed network figure is sent to connect with this main controlled node it is all from node.
B8, the method as described in B6, wherein the mission bit stream is the mission bit stream of deep learning task；The task
Information includes: for executing the number of nodes of deep learning task, deep learning subtask type, all types of subtask numbers
Amount.
B9, the method as described in B8, wherein the mission bit stream according to input is sent to one or more from node
Task start instructs
From all number of nodes for being selected from node and being used to execute deep learning task being connect with this main controlled node
It is comparable from node；
According to deep learning subtask type and all types of subtask quantity, determines and appoint what is respectively started from node
Business；
To the slave node of each selection send in starting from node for the task corresponding task start instruction.
The embodiment of the present invention also discloses C10, a kind of meshed network self-organizing device, wherein the device is deployed in point
On the slave node of cloth cluster, comprising:
Communication unit, the task start instruction sent suitable for receiving node self-organization of network server；
Port selection unit, suitable for choosing the port numbers for being used for the task；
The communication unit is further adapted for return to where the present apparatus from the host name of node and the port numbers of selection described
Meshed network hoc service device, so that the meshed network hoc service device is according to respectively starting from node for task and each section
The host name and port numbers that spot net self-organizing device returns generate meshed network figure；And it is suitable for receiving the meshed network
The meshed network figure that hoc service device is sent.
C111, the device as described in C110, wherein
The port selection unit, suitable for where the present apparatus from the current unappropriated port numbers of node, at random
Choose a port number.
C112, the device as described in C110, wherein
The communication unit obtains asking for meshed network figure suitable for periodically sending to the meshed network hoc service device
It asks, receives the meshed network figure that the meshed network hoc service device is returned according to the request；And/or receive the node
The meshed network figure that self-organization of network server actively issues.
C113, the device as described in C110, wherein
The communication unit is further adapted for the meshed network figure sent according to the meshed network hoc service device, with this
Other one or more in meshed network figure establish connection from the meshed network self-organizing device on node.
C114, the device as described in C110, wherein
The task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask includes:
The subtask parameter server and/or the subtask worker.
The embodiment of the present invention also discloses D15, a kind of meshed network hoc service device, wherein the server disposition
On the main controlled node of distributed type assemblies, comprising:
Communication unit, suitable for the mission bit stream according to input, to one or more from node on meshed network self-organizing
Device sends task start instruction；Receive host name and port numbers that each meshed network self-organizing device returns；
Meshed network figure generation unit, suitable for being returned according to the task and each meshed network self-organizing device that respectively start from node
The host name and port numbers returned generate meshed network figure；
The communication unit is further adapted for for the meshed network figure being sent to one or more from the meshed network on node
Self-organizing device.
D16, the server as described in D15, wherein
The communication unit, suitable for receiving the acquisition node net sent from the meshed network self-organizing device on node
When the request of network figure, the meshed network figure is sent to this from the meshed network self-organizing device on node；And/or by institute
It states meshed network figure and is sent to all meshed network self-organizings from node connecting with the main controlled node where book server
Device
D17, the server as described in D15, wherein the mission bit stream is the mission bit stream of deep learning task；It is described
Mission bit stream includes: for executing the number of nodes of deep learning task, deep learning subtask type, all types of subtasks
Quantity.
D18, the server as described in D17, wherein the server further include:
Scheduling unit, suitable for from connect with the main controlled node where book server it is all from node selection be used for hold
The number of nodes of row deep learning task is comparable from node；According to deep learning subtask type and all types of subtask numbers
Amount is determined in respectively starting from node for task；
The communication unit, suitable for the meshed network self-organizing device on the slave node to selection send with this from node
The corresponding task start instruction of the task of upper starting.
The embodiment of the present invention also discloses E19, a kind of meshed network self-organizing system, wherein the system includes one
Or multiple meshed network self-organizing devices as described in any one of C10-C14 and such as any one of claim D15-D18 institute
The meshed network hoc service device stated.
Claims (17)
1. a kind of meshed network self-organizing method, wherein this method comprises:
Receive the task start instruction that main controlled node is sent；
Choose the port numbers for being used for the task；
This is returned into the main controlled node from the host name of node and the port numbers of selection, so that the main controlled node is according to each
The host name and port numbers of the task and return that start from node generate meshed network figure；
Receive the meshed network figure that the main controlled node is sent；
Described choose include: for the port numbers of the task
From this from node currently unappropriated port numbers, a port number is randomly selected.
2. the method for claim 1, wherein the meshed network figure for receiving the main controlled node transmission includes:
The request for obtaining meshed network figure is periodically sent to the main controlled node, is received the main controlled node and is returned according to the request
Meshed network figure；
And/or
Receive the meshed network figure that the main controlled node actively issues.
3. the method for claim 1, wherein this method further include:
According to the meshed network figure that the main controlled node is sent, built with other one or more in the meshed network figure from node
Vertical connection.
4. the method for claim 1, wherein
The task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask includes:
The subtask parameter server and/or the subtask worker.
5. a kind of meshed network self-organizing method, wherein this method comprises:
According to the mission bit stream of input, task start instruction is sent to one or more from node；
Receive the host name and port numbers respectively returned from node；The port numbers are respectively currently not occupied from node from node from this
One randomly selected out in port numbers；
According to the host name and port numbers of the task and return that respectively start from node, meshed network figure is generated；
The meshed network figure is sent to one or more from node.
6. method as claimed in claim 5, wherein described that the meshed network figure is sent to one or more from node packet
It includes:
When receiving the request of the acquisition meshed network figure sent from node, the meshed network figure is sent to this from section
Point；
And/or
The meshed network figure is sent to connect with this main controlled node it is all from node.
7. method as claimed in claim 5, wherein the mission bit stream is the mission bit stream of deep learning task；Described
Information of being engaged in includes: for executing the number of nodes of deep learning task, deep learning subtask type, all types of subtask numbers
Amount.
8. the method for claim 7, wherein the mission bit stream according to input is sent out to one or more from node
Send task start instruction include:
From and all selections from node that connect of this main controlled node be used for that execute the number of nodes of deep learning task suitable
Slave node；
According to deep learning subtask type and all types of subtask quantity, determine in respectively starting from node for task；
To the slave node of each selection send in starting from node for the task corresponding task start instruction.
9. a kind of meshed network self-organizing device, wherein the device is deployed on the slave node of distributed type assemblies, comprising:
Communication unit, the task start instruction sent suitable for receiving node self-organization of network server；
Port selection unit, suitable for choosing the port numbers for being used for the task；
The communication unit is further adapted for that the node will be returned to from the host name of node and the port numbers of selection where the present apparatus
Self-organization of network server, so that the meshed network hoc service device is according to the task and each node net respectively started from node
The host name and port numbers that network self-organizing device returns generate meshed network figure；And it is suitable for receiving the meshed network from group
Knit the meshed network figure of server transmission；
The port selection unit, suitable for from the current unappropriated port numbers of node, being randomly selected where the present apparatus
A port number.
10. device as claimed in claim 9, wherein
The communication unit, suitable for periodically sending the request for obtaining meshed network figure to the meshed network hoc service device,
Receive the meshed network figure that the meshed network hoc service device is returned according to the request；And/or receive the meshed network
The meshed network figure that hoc service device actively issues.
11. device as claimed in claim 9, wherein
The communication unit is further adapted for the meshed network figure sent according to the meshed network hoc service device, with the node
Other one or more in network establish connection from the meshed network self-organizing device on node.
12. device as claimed in claim 9, wherein
The task start instruction is the enabled instruction of deep learning subtask；The deep learning subtask includes:
The subtask parameter server and/or the subtask worker.
13. a kind of meshed network hoc service device, wherein the server disposition is on the main controlled node of distributed type assemblies, packet
It includes:
Communication unit, suitable for the mission bit stream according to input, to one or more from node on meshed network self-organizing device
Send task start instruction；Receive host name and port numbers that each meshed network self-organizing device returns；The port numbers are each
Randomly selected out from node currently unappropriated port numbers from this one of meshed network self-organizing device；
Meshed network figure generation unit, suitable for what is returned according to the task and each meshed network self-organizing device that respectively start from node
Host name and port numbers generate meshed network figure；
The communication unit is further adapted for for the meshed network figure being sent to one or more from the meshed network on node from group
Knit device.
14. server as claimed in claim 13, wherein
The communication unit, suitable for receiving the acquisition meshed network figure sent from the meshed network self-organizing device on node
Request when, the meshed network figure is sent to this from the meshed network self-organizing device on node；And/or by the section
Spot net figure is sent to all meshed network self-organizing devices from node connecting with the main controlled node where book server.
15. server as claimed in claim 13, wherein the mission bit stream is the mission bit stream of deep learning task；Institute
Stating mission bit stream includes: to appoint for executing the number of nodes of deep learning task, deep learning subtask type, all types of sons
Business quantity.
16. server as claimed in claim 15, wherein the server further include:
Scheduling unit, suitable for from all selections from node being connect with the main controlled node where book server and for executing depth
The number of nodes for spending learning tasks is comparable from node；According to deep learning subtask type and all types of subtask quantity,
It determines in respectively starting from node for task；
The communication unit sends suitable for the meshed network self-organizing device on the slave node to selection and is opened from node at this
The corresponding task start instruction of dynamic task.
17. a kind of meshed network self-organizing system, wherein the system includes one or more such as any one of claim 9-12
The meshed network self-organizing device and the meshed network hoc service device as described in any one of claim 13-16.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610963725.XA CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610963725.XA CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106411605A CN106411605A (en) | 2017-02-15 |
| CN106411605B true CN106411605B (en) | 2019-05-31 |
Family
ID=58014284
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610963725.XA Active CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106411605B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109614232B (en) * | 2018-12-07 | 2021-02-19 | 网易（杭州）网络有限公司 | Task processing method and device, storage medium and electronic device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101888331A (en) * | 2009-05-13 | 2010-11-17 | 阿瓦亚公司 | Be used to provide the method and apparatus of the quick rerouting of unicast packet |
| CN104158747A (en) * | 2013-05-14 | 2014-11-19 | 中兴通讯股份有限公司 | Network topology discovery method and system |
| CN105391580A (en) * | 2015-11-27 | 2016-03-09 | 上海卫星工程研究所 | Network model description method applicable to SpW/SpF |
| CN105721318A (en) * | 2016-02-29 | 2016-06-29 | 华为技术有限公司 | Method and device for discovering network topology in software defined network SDN (Software Defined Network) |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7391771B2 (en) * | 2004-01-23 | 2008-06-24 | Metro Packet Systems Inc. | Method of sending information through a tree and ring topology of a network system |
-
2016
- 2016-10-28 CN CN201610963725.XA patent/CN106411605B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101888331A (en) * | 2009-05-13 | 2010-11-17 | 阿瓦亚公司 | Be used to provide the method and apparatus of the quick rerouting of unicast packet |
| CN104158747A (en) * | 2013-05-14 | 2014-11-19 | 中兴通讯股份有限公司 | Network topology discovery method and system |
| CN105391580A (en) * | 2015-11-27 | 2016-03-09 | 上海卫星工程研究所 | Network model description method applicable to SpW/SpF |
| CN105721318A (en) * | 2016-02-29 | 2016-06-29 | 华为技术有限公司 | Method and device for discovering network topology in software defined network SDN (Software Defined Network) |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106411605A (en) | 2017-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105389191B (en) | A kind of method for upgrading software based on LAN, device and system | |
| EP2710470B1 (en) | Extensible centralized dynamic resource distribution in a clustered data grid | |
| US10462000B2 (en) | Micro-services for configuring switches | |
| CN105100259B (en) | A kind of distributed timing task executing method and system | |
| US8412810B1 (en) | Provisioning and managing a cluster deployed on a cloud | |
| CN105229973B (en) | The composition of network and the visualization system of action | |
| US11422844B1 (en) | Client-specified network interface configuration for serverless container management service | |
| CN106529682A (en) | Method and apparatus for processing deep learning task in big-data cluster | |
| Schneider et al. | Self-learning multi-objective service coordination using deep reinforcement learning | |
| CN107515786A (en) | Resource allocation methods, master device, from device and distributed computing system | |
| CN106569878A (en) | Distributed timing task management method and system | |
| CN108234242A (en) | A kind of method for testing pressure and device based on stream | |
| CN106411605B (en) | A kind of meshed network self-organizing method, device, server and system | |
| CN113382077B (en) | Micro-service scheduling method, micro-service scheduling device, computer equipment and storage medium | |
| CN111163140A (en) | Method, apparatus and computer readable storage medium for resource acquisition and allocation | |
| CN110290228A (en) | A kind of internet protocol address distribution method and device | |
| CN110163554A (en) | Operation method, device, server and the storage medium of workflow | |
| KR101899298B1 (en) | Processing system for automation of cloud and process operating based virtual environment and processing method thereof | |
| Abreu et al. | Resilient service chains through smart replication | |
| JP6707198B2 (en) | Distributed arrangement apparatus, distributed arrangement system, and distributed arrangement method | |
| JP2020126487A (en) | Parallel processor apparatus, data transfer destination determining method, and data transfer destination determining program | |
| CN109710878A (en) | Data interactive method and server | |
| Saad et al. | Designing and implementing a cloud-hosted SaaS for data movement and sharing with SlapOS | |
| CN114020408A (en) | Task fragment configuration method and device, equipment and storage medium | |
| CN109032786A (en) | Jenkins continuous integrating cluster, APP packaging method and server |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220725 Address after: Room 801, 8th floor, No. 104, floors 1-19, building 2, yard 6, Jiuxianqiao Road, Chaoyang District, Beijing 100015 Patentee after: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Address before: 100088 room 112, block D, 28 new street, new street, Xicheng District, Beijing (Desheng Park) Patentee before: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Patentee before: Qizhi software (Beijing) Co.,Ltd. |
|
| TR01 | Transfer of patent right |