CN109995875A - Virtualization data distributes elastomeric network system - Google Patents
Virtualization data distributes elastomeric network system Download PDFInfo
- Publication number
- CN109995875A CN109995875A CN201910289822.9A CN201910289822A CN109995875A CN 109995875 A CN109995875 A CN 109995875A CN 201910289822 A CN201910289822 A CN 201910289822A CN 109995875 A CN109995875 A CN 109995875A
- Authority
- CN
- China
- Prior art keywords
- network
- data
- virtualization
- distributes
- network system
- 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.)
- Withdrawn
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
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
Abstract
The present invention provides a kind of virtualization data distribution elastomeric network system, virtualization process is carried out to each Internet resources being arranged in the whole network, SDN network is established in network layer, when each Internet resources in SDN network carry out data transmission, uses the DDS publish/subscribe mode of multiregion system in transport layer.The present invention is improved in network layer, transport layer for current coal industry field and enterprise network system, maximize the service efficiency of link, the efficiency of transmission and response real-time for improving business-critical, ensure that the service quality of data transmission, realize the real-time transmission and distribution of critical data flow.
Description
Technical field
The present invention relates to a kind of mining virtualization datas of software definition to distribute elastomeric network system.
Background technique
In current coal industry field and enterprise network system, in the communication of data and the link use aspect of flow, because of net
The unreasonable distribution of network resource, causes business-critical that cannot send and transmit in time, is handled on an equal basis with general information.?
When being communicated, and because cannot specification distribution efficiency, cause critical data to be unable to real-time transmission and distribution.
Summary of the invention
The present invention provides a kind of virtualization datas to distribute elastomeric network system, for solving above-mentioned critical data flow
The problem of transmission and distribution, it is described that its technical solution is as follows:
A kind of virtualization data distribution elastomeric network system, carries out at virtualization each Internet resources being arranged in the whole network
Reason, establishes SDN network in network layer, when each Internet resources in SDN network carry out data transmission, uses multiple domain in transport layer
The DDS publish/subscribe mode of system.
Communication protocol in SDN network system uses OpenFlow agreement, real between the network equipment and SDN controller
It is existing.
SDN network based on OpenFlow can dispose in existing network, including physical network, virtual network and
The hybrid network environment of multiple commercial vendors equipment.
To the distribution of the dynamic high-efficiency of flow by virtualization floating resources dispatching method in SDN network, this method is by network
In flow virtualization after, being divided into general data flow and has the special data flow of QoS demand.
Virtualization floating resources dispatching method realizes different routings to general data flow and special data flow respectively
Algorithm allows the routing of general data flow to avoid special data flow by link weight realization.
Virtualization floating resources dispatching method provides special data flow the routing algorithm under more QoS index constraints, real
The dynamic of now all special critical data streams adjusts, and realizes the high-speed transfer of critical data flow.
It is excessively high to there is link load in network, when especially more serious packet loss phenomenon occurs in network, virtualizes floating resources
Dispatching method is using the data grouping of discarding part common data stream come the QoS demand of Logistics networks system.
DDS publish/subscribe mode is data-centered, each back end (each Internet resources) on a distribution network with
Publication or the mode subscribed to transmit data, and node is publisher or subscriber, or is both publisher and subscriber.
DDS includes DCPS and two layers of DLRL, and DCPS is a kind of data-centered information exchange model, the reality of DCPS
Body includes domain, domain participant, data writer, publisher, data reader, subscriber, theme；The DLRL layers of base established in DCPS
On plinth, allow service being simply integrated into application layer, for recombination data automatic after having data update, and notifies institute
There is subscriber to update local data.
Multiregion system adds new domain and has no effect to original domain in current system.
To virtualization data distribution elastomeric network system detect when, using Wireshark to the flow in network into
Row packet capturing analysis, passes through the bandwidth performance of TCP and UDP in iperf performance testing tool test network, the performance test work
Tool can report the bandwidth of network flow, delay jitter and data-bag lost state.
The present invention is improved in network layer, transport layer for current coal industry field and enterprise network system.With with
Lower advantage:
1) IP network is realized by carrying out virtualization process to Internet resources, carrying out software function virtualization process to network element
The rationalization distribution of link circuit resource, floating resources, to maximize the service efficiency of link, improves the biography of business-critical in network
Defeated efficiency and response real-time.
2) it is realized by using the data communication normal form of publish-subscribe model and based on hard real time transmission control technology soft
Part communication construction redefines transmission data format, transmission channel, data sharing spatial cache etc., realizes one
The efficient Data distributing system of loose coupling is covered, to ensure that the service quality of data transmission, realizes critical data
The real-time transmission and distribution of flow.
Detailed description of the invention
Fig. 1 is SDN technical principle schematic diagram；
Fig. 2 is client end/server end model schematic (C is client, and S is server)；
Fig. 3 is the system structure diagram of message queue model；
Fig. 4 is the configuration diagram of publish/subscribe model；
Fig. 5 is the schematic diagram for issuing and subscribing to flexible relationship in typical publish/subscribe model；
Fig. 6 is DDS data publication and a pair of process schematic for the moment of subscription transmission；
Fig. 7 is the DLRL layer schematic diagram that DDS hierarchical structure includes DCPS layers and its upper layer；
Fig. 8 is the multiregion system schematic diagram of DDS；
Fig. 9 be the domain inside the message of the privately owned subsystem in multiregion system uses and can be used external domain as
Common interface schematic diagram；
Figure 10 is that multiregion system establishes virtual data network schematic diagram in the same physical network；
Figure 11 is the realization schematic diagram for subscribing to model and data reader；
Figure 12 is the industrial network design diagram in coal mine application scenarios.
Specific embodiment
Virtualization data provided by the invention distributes elastomeric network system, carries out to each Internet resources being arranged in the whole network
Virtualization process establishes SDN network in network layer, when each Internet resources in SDN network carry out data transmission, in transport layer
Using the DDS publish/subscribe mode of multiregion system.Related introduction is carried out to SDN and DDS below.
Elastic Internet resources distribution technique realization means in SDN:
Software definition is that network (SDN) technology is a kind of network architecture that network-control is mutually separated with data forwarding,
And network control function has programmable features, can be configured flexibly network function by software program.Current network
Equipment, network control function (including management and routing decision etc.) and network equipment close-coupled, software defined network it is basic
Thinking is to be stripped out network control function, and using software realization, such mode can be by network control function concentrate tube
Reason, enables the network equipment to be absorbed in a kind of function, convenient for the demand of network upgrade and the various new business of adaptation.
Software defined network architecture is as shown in Figure 1.Network-control decision concentrates on software-based SDN control function
In energy entity.SDN network is virtualized by Internet resources, customization network is realized, by managing and cooperateing with control concentratedly
System, realizes the efficient utilization of Internet resources, by open network control layer and data forwarding interface, realizes new technology
Quickly application.It can be obtained by using the network management quotient and various users of SDN technical network unrelated with network equipment vendor
(current network equipment function binding, equipment user can not change network equipment function to network control capabilities, it is necessary to network
Equipment vendor participates in that related transformation could be implemented).In this way, network design and operation will all simplify.SDN equally can be with
Simplify the design of the network equipment, in SDN, the network equipment no longer needs " to understand " and " execution " thousands of network protocol,
Only need to be implemented the instruction that SDN controller issues.
SDN by realize from equipment vendor programs to operator program network system management transformation of the functions, equipment may be programmed to
Network Programmable transformation, network and application are by Seamless integration-.The basic thought and main feature of SDN include:
Network-control face is separated with data forwarding face, supports third party's net control device by standard agreement to data
Forwarding device is controlled.
The network-control face function of logical centralization improves the flexibility of the network-controls such as routing management, accelerates service fulfillment
Speed simplifies O&M.
Data forwarding face Device-General, a variety of route exchange devices can share general hardware platform.
Network control software is programmable, meets user by programmable software and customizes demand.
SDN network provides a series of API (application programming interfaces).These API support the service of many general network, packet
Include routing, multicast, security protection, access control, Bandwidth Management, traffic engineering, QoS guarantee, processing memory optimizing, energy
Management and various tactical managements.For example, SDN can be very easily to there is Wireless Heterogeneous Networks to be managed and optimize.
Crucial communication protocol is OpenFlow agreement in SDN network system, is controlled in first SDN network of industry
Standard communication interface agreement between layer and data forwarding.OpenFlow agreement is real between the network equipment and SDN controller
It is existing.OpenFlow identifies various network flows using the concept of stream, is controlled by pre-defined, programmable regular convection current
System.The network equipment of OpenFlow is supported to carry out data forwarding according to flow table.The generation of the not responsible flow table of network device itself and
Maintenance, these work are completed by external, operational network operating system on computers controller completely.Controller passes through
Standard agreement is by flow table issuance to the network equipment.In OpenFlow1.0, the packet header domain in flow table define including port numbers,
12 fields of virtual LAN ID number and L2/L3/L4 information.Since each packet header domain can carry out wildcard, network fortune
Battalion administrative staff can according to need the stream for determining to use which kind of granularity.Since OpenFlow allows based on stream predetermined
Rule carries out network-control, thus the SDN network based on OpenFlow can be real-time according to factors such as application, user and sessions
Adjust Internet resources.
OpenFlow agreement is the SDN network control protocol of current comparative maturity, obtains the support of multiple commercial vendors.Base
It can be disposed in existing network in the SDN network of OpenFlow, including physical network, virtual network and multiple commercial vendors are set
Standby hybrid network environment.The network equipment of multiple commercial vendors has been able to support OpenFlow agreement at present, these equipment are in portion
The data forwarding of OpenFlow can be supported in administration, can also be worked with traditional data pass-through mode.
Realize that the dynamic high-efficiency distribution of flow relies primarily on virtualization floating resources dispatching technique in SDN network, by net
After flow virtualization in network, it is divided into common discharge and has the special flow of QoS demand.Virtualize floating resources dispatching method
It can be summarized as routing algorithm different with special data flow design to general data flow respectively.Pass through software definition
Mode cleverly designs link weight, and the routing of common discharge is allowed to avoid special data flow as far as possible, reduces common discharge burst
Property influence to special flow QoS index, realization reliably distributes the link circuit resource of special business-critical.Meanwhile to special number
The routing algorithm under more QoS index constraints is provided according to flow, the dynamic adjustment of all special critical data streams is realized, realizes and close
The high-speed transfer of key data flow improves the handling capacity of network, reduces average request response delay.In order to guarantee spy as far as possible
The reliable and effective transmission of different data traffic, if link load occur excessively high for network, especially it is existing to there is more serious packet loss for network
As when, the data grouping that network can abandon part common data stream carrys out the QoS demand of Logistics networks system.
Technology for Ensuring QoS realization means in DDS
In industrial Internet of things system, how to obtain data in real time is very crucial problem.In this kind of real-time application
In want to accomplish " obtaining correct data (3R) in correct time, correct place ", it is clear that this requirement to system is very
It is high.However traditional C/S model the problems such as not can be well solved real-time, reliability and dynamic extensibility, therefore it is uncomfortable
Close this kind of real-time system of industrial application.
The shortcomings that generation of publish/subscribe (Publish/Subscribe) model compensates for C/S model just, is C/S
The strong supplement of model, especially data-centered publish/subscribe model, it can provide reliable data point in real time
Hair service.
There are three types of usually used data distribution models: client end/server end (C/S) model, message queue (MQ) mould
Type and publish/subscribe (P/S) model.
Client end/server end (C/S) model
Client end/server end model is usually made of a server and multiple client, as shown in Figure 2.The model
The characteristics of be: client directly establishes connection with server end, backward server end request service and waiting for server end
Response does corresponding processing after the request of received server-side to client, and processing result is returned to client, leads to
It crosses this mode and provides service to client.In the process, the promoter of communication is client.Common CORBA, RMI,
DCOM, EJB etc. are exactly to use C/S model.
As can be seen that this is an one-to-many architecture in the characteristics of C/S model.In the model, client with
The degree of coupling of server end is high.Therefore, the scalability of system, flexibility, robustness are lower.If client access is clothes
The centralized resources (such as database) being engaged on device end, then the model is applicable.But in distributed real-time systems, with
The increase of client node, the increased workload of server end, the response time increases, to performance bottleneck occur.And
In this model, when server end breaks down, whole system " will paralyse ", that is, there are problems that single point failure.Therefore,
C/S model can not meet very well as intelligent mine system (when especially including hard real time control system) is this kind of to real-time
Property, the very high real-time system of reliability requirement.
Message queue (MQ) model
Application program in different network hardware platforms, different operating system or even different network protocols it
Between sometimes for transmission message, the requirement that at this moment application program transmit data be the content of passed message it is reliable with can restore
If (accident occurs), and other side can be immediately transmitted to immediately by being not required for data.So this just needs a kind of message-oriented
Data distribution model, i.e. message queue model, as shown in Figure 3.
In this model, allow each application program to operate on different nodes, and allow different message
Format.Message sender sends a message to message queue in a queue, and addressee can check the message at any time
Queue.Message queue and present e-mail system are very similar, can guarantee the quality of message transmission, but can not know receipts
Whether part people reads the message.Since the model needs not distinguish between server end and client, present in C/S model
Performance bottleneck problem and scalability issues are just not present.But the TRANSFER MODEL is not data-centered, and
It is not related to QoS, so the model is also not suitable for real-time system.
Publish/subscribe (P/S) model
Publish/subscribe model realizes application program as shown in figure 4, being the data distribution model haveing excellent performance
Between efficiently distribute data.In this model, node only needs to subscribe to the data interested to it, or issues its institute
The data that can be provided.The former is known as subscriber (Subscriber), and the latter is known as publisher (Publisher), sends out in the model
Communication between cloth person and subscriber is entirely to be completed by middleware.Between the two without knowing mutual presence, work as hair
When the person of sending sends data, the address for knowing subscriber and quantity are not needed, and need to only be concerned about the certain number transmitted required for it
According to；When subscriber receives information, without the address and quantity for knowing publisher, and need to only know received desired by it
Specific data.
Two kinds of this model and front the difference is that, in systems permit there are multiple publishers and subscribers, and
There is no stringent to define by role.At any time, be between publisher and subscriber it is transparent, they are without knowing other side
Address and presence, each node in system is only arbitrary, dynamically, on demand serve as publisher and subscriber role.
Just because of this transparency, publish/subscribe system realizes being dynamically added or exiting for each publisher/subscriber.It is typical
Publish/subscribe model it is as shown in Figure 5.
Compared with two kinds of models above, it between each node is weak coupling that P/S model biggest advantage, which is in system,
Relationship, which achieves the asynchronous independence between each node.When being in particular in, in terms of sky, stream three:
(1) Time Decoupling (TimeDecoupling): i.e. publisher and subscriber are not necessarily to while being in ready state.?
When publisher issues data, subscriber may be without online；After publisher is offline, subscriber is also possible to receive corresponding
The data that publisher is issued.The function of this asynchronous transmission is provided by middleware.
(2) it spatially decoupled (SpaceDecoupling): is not necessarily to know the address of other side i.e. between publisher and subscriber
And presence.Publisher only needs to issue the data that it can be provided, and subscriber only needs to receive the data interested to it,
There is no direct connection between publisher and subscriber, do not need to know the information such as address and the port numbers of other side in advance.
(3) data flow decoupling (FlowDecoupling): i.e. the host process of publisher and subscriber be it is asynchronous, be not required to
Block.When issuing data, itself need not block publisher；Subscriber need not also incite somebody to action when receiving interested data
Itself obstruction, but other operations can be carried out simultaneously.Exactly these characteristics of publish/subscribe model eliminate each logical
Interdepending between reliable body substantially increases the flexibility and scalability of system, provides asynchronous transmission function.And
This expansible characteristic enables the model preferably to meet the requirements of distributed real-time systems than other models.
Existing publish/subscribe system has: CORBA NotificationService, JMS, JEDI, TIB/
Rendezvous, Siena, Gryphon, Scribe etc..But these systems all lack the QoS support of application layer, therefore these are
System can not meet the requirement in terms of real-time and reliability well.It is wanted to meet real-time and the reliability of real-time system
It asks, OMG tissue has issued the specification based on data-centered publish/subscribe (DCPS) traffic model, i.e. DDS.The specification
Interface and the behavior that standardization is defined for DDS middleware propose data-centered publish/subscribe mechanism, provide
One and platform-independent data distribution model.In addition, DDS specification also define a large amount of qos policy, allow DDS very
System resource is configured and utilized well, coordinates the balance between predictability and execution efficiency, supports data flow complicated and changeable
Demand.
Data distributing DDS is a specification about publish/subscribe model in distributed real-time systems, this rule
Model defines a data-centered publish/subscribe model, provides a middleware framework independently of platform, is
Data publication, transmitting and received interface and behavior provide unified standard in real-time system.It allows application program real-time
Its information that can be provided is issued on ground, and subscribes to required information.In addition to this, DDS also supports many QoS attributes, such as different
Step, loose coupling, the distribution of real-time authentic data etc..The purpose of DDS specification is effective publication of data in simplified distributed system,
It is suitable for performance requirement height, predictable strong real time critical tasks in areas.The data publication of DDS and subscription simple one
It is as shown in Figure 6 to one transmission process.It can satisfy between equipment come only DDS technology as can be seen from Figure 6 and equipment and remote
Hard real time control signal transmission between process control facility.
DDS service is data-centered, and each node transmits data in a manner of issuing or subscribe on a distribution network.
Distributed network does not have center, because without because center by destroying due to cause the collapse of system entirety.Node can be hair
Cloth person or subscriber, or be both publisher and subscriber.The loosely coupled architecture of DDS point-to-point transmission, with traditional mapping side
It is compared in formula using the close coupling communication structure of C/S model, can preferably meet the spirit that coal mine operation system transmits data
The requirement of activity and reliability.
DDS specification can be divided into DCPS (Data-CentricPublish-Subscribe) and DLRL
(DataLocalReconstructionLayer) two layers, as shown in Figure 7.Wherein, the traffic model that DDS is used is with data
Centered on publish/subscribe DCPS model.DCPS is a kind of purely data-centered information exchange model, the reality of DCPS
Body generally comprises: domain (Domain), domain participant (DomainParticipant), data writer (DataWriter), publication
Person (Publisher), data reader (DataReader), subscriber (Subscriber), theme (Topic).Based on the overall situation
The node of the concept of data space, the feedback in all pairs of global spaces can access, the data object in network
It is identified with domain or theme.The node that data are provided to global data space is " publisher ", needs to obtain number from global space
According to being " subscriber ", the information of publish/subscribe is identified by " theme ".Each node is with complete independence and certainly
Main property, be all logically without master-slave relationship, between node and node equity relationship, communication mode can be it is one-to-one, one
To more, many-one and multi-to-multi etc..Publisher issues data in the form broadcasted, subscriber judge by comparing theme be
No is the data that oneself is needed.
In addition, the DCPS layers of support additionally provided to a variety of QoS (Quality of Service, QoS) strategy.QoS is
A kind of network transmission strategy, network transmission quality behavior required for application program is formulated, QoS service realize that this behavior is wanted
It asks, meets the needs of client is to communication quality as far as possible.Each node has a qos policy of itself, and each pair of publisher and
It can establish independent QoS agreement between subscriber.Node establishes connection under the control of quality of service policy, automatic to find
With Configuration network parameter.
DLRL (Data Local Reconstruction Layer) layer is located at high level, establishes on the basis of DCPS
On, it is mono- optional level of DDS, allows service being simply integrated into application layer.Number is recombinated automatically after having data update
According to, and all subscribers is notified to update local data.
DDS is realized the data exchange ways independently of platform, can be mapped using Interface Definition Language description service
To a variety of platform specifics and programming language.Therefore DDS can support different processor architectures, programming language and operation
The combination of system.Currently, DDS solution can be used a variety of programming languages (mainly C, C++ and Java), and for a variety of
Operating system, such as VxWorks, QNX, Lynx, Windows and Unix/Linux.For DDS technology, relevant some keys
Concept is described below.
Send and receive data: this mode is helpful to being isolated and optimizing the communication that those share public interest.One
It can also can be made of multiple domains with only one domain in a distributed system, the former is known as single domain system；The latter is known as multiple domain
System (as shown in the figures such as Fig. 8,9,10).In single domain system, all participants are communicated in this domain.And
In multiregion system, developer can need to set multiple domains according to system.When a specific data instance is sent out in one domain
When cloth, the subscriber in another domain will be unable to receive the data instance.As shown in figure 8, multiregion system, which provides, more to be had
The data independence of effect, what it is in Domain A transmission is Command/Control data, and what is transmitted in Domain B is
Status data, developer can set different domains according to different functions.
Another advantage of multiregion system is: good solution is provided to add new function in current system,
Developer only needs to add a new domain in systems, and the domain newly added has no effect to original domain.
Domain participant (Domain Participant): DDS application program movable reality in domain can be shown by being one
Body, it is the founder and manager of DCPS entity (data writer, publisher etc.).In addition, developer can also pass through domain
Participant is the corresponding QoS parameter of all entity sets in the domain of dependence.Data writer is the typed access of publisher
Device, each data writer are only associated with a specific theme, therefore only have a kind of data type.Application program passes through data
The particular type of interface of writer issues the data instance of associated theme.Data writer is responsible for passing data to publisher,
And data distribution is exactly the task of publisher, publisher can according to the QoS of itself QoS and corresponding data writer come
Distribute data.
Data reader (Data Reader) and is passed data to for obtaining the data as received by subscriber
The DCPS entity of application program, each data reader are bound to a specific theme, that is to say, that each data reader is only
It is concerned about the data type interested to it.Subscriber is responsible for receiving the data from publisher, then passes the data received
To corresponding data reader, so that application program be made to obtain the data interested to it.Subscriber is by data reader's group in fact
It is woven in together, and is created by domain participant in itself.Figure 11 is shown with subscribing to relationship (its between relevant each entity
In, data reader can obtain the data ordered by it, both modes by Listener and Waitset two ways
Difference will make narration in detail in " informing mechanism ").Data reader provides great flexibility for developer, can be with
Directly by calling function to obtain data from DDS middleware.
Theme (Topic): being that basic tie point is provided between publisher and subscriber, the publication on a node
Person's theme has to match with subscriber's theme on another node, if theme mismatches, publisher and subscriber
Between cannot communicate.One theme is made of subject name (TopicName) and type of theme (TopicType).Subject
Title be one can in domain unique identification theme character string, and type of theme is included in the definition of the data in theme.It is main
Topic must be unique in a specific domain.In DDS frame, if two themes have different subject names but phase
Same type of theme, then it is assumed that the two themes are different.In addition, each theme can also be associated with corresponding qos policy.
Informing mechanism: data reader can obtain the data that it is subscribed to by following two approach.One is based on prison
Listen device (Listener-based).Which provides a kind of asynchronous mechanism to obtain data, specially there is one in the application
A separate threads are responsible for monitoring, and have data available arrival using call-back manner notice application program.Due to using thread,
Therefore this mode will not block the operation of primary application program.Another kind is based on waiting set (Waitset-based).It should
Mode provides a kind of synchronization mechanism, and application program will always be in blocked state until specified condition meets.
In order to verify the feasibility that this research institute mentions virtualization elastic data distributing network system, simulation system used below
System builds, the research method of system testing verifies the performance of used technology.
1) network environment of a set of simulation coal production environment is built (such as according to the actual conditions of network system in coal mine
Shown in Figure 12) run this system.
2) stability of system is tested according to the technical standard of system and technical indicator, test meets system and refers to
Mark.
What this research project proposed is the mining virtualization elastic data distributing network system scheme of software definition.The party
Case can be divided into two resource management modules from structure: first be path optimization's algorithm QoS flow policy module.If
Network congestion phenomenon can disappear after the redistributing of passage path, and all flows in network guarantee normal transmission, then side
Case management terminates.Second resource management module is proposed to improve the flexibility to network resource management.It is
A kind of scheme is redistributed after resource there is no guarantee that starting to execute under conditions of network performance, and proposition passes through Internet resources
The Managed Solution of virtualization realizes the isolation to network topology, bandwidth and flow, reduces the influence to high precedence stream, with
The performance of Logistics networks.Since the main content of this chapter is to propose the virtual management scheme of Internet resources, for the mesh of experiment
To verify performance in resource management scheme, pass through the bandwidth threshold certain for link setting in the design of test macro
Value guarantees to trigger the request life for executing elastomeric network resource virtualizing by event in the case where path computing failure
It enables, realizes the virtual management to network flow resource.
This test process selects physical equipment as experiment test platform, has both been able to verify that QoS flow policy module path optimization
Scheme, and Internet resources can be virtualized and be tested, to verify the scalability and stability of resource management scheme.?
Use 3 OpenFlow interchangers and a SDN controller and 6 hosts respectively as the client of network in experiment altogether
End and server-side.The concrete configuration of equipment needed for testing is as shown in table 4-3 and table 4-4.
The gigabit Ethernet of tetra- interface of DELL INTE L1350 T4QP is mounted on association M43620 model Pc herein
Card is to extend network card interface construction OpenFlow interchanger.It is installed on this machine by innovative development research institute, Ericsson simultaneously
1.4 software switch of OpenFlow of publication, the interchanger are a kind of software switch based on user's space.It is sent out with ONF
Based on 1.4 agreement of OpenFlow of cloth, 1.4 agreement of OpenFlow is supported.Running environment Ubuntu 14.04.
Use Dell Latitude E6530 as the carrier of SDN controller, onos 1.9.0 version is installed on it
As controller software system, running environment Ubuntu 12.04.
Application layer data is packaged using the DDS Data distributing technology of independent research, and needs to determine according to experiment
Adopted corresponding data transmission procedure QoS priority.
This project carries out packet capturing analysis to the flow in network using Wireshark, passes through iperf performance testing tool
The bandwidth performance of TCP and UDP in test network, the performance testing tool can report the bandwidth of network flow, delay jitter
With data-bag lost state.
Controller is by distributing different isolation of the ID realization to flow of joining the team for different flow, due to each queue pair
It answers each virtualization network all to set corresponding occupied bandwidth, therefore virtualizes network and ensure that the transmission of flow.This survey
The purpose of examination experiment is mainly to verify the Resource Allocation Formula of network virtualization, uses three OpenFlow exchange mechanisms herein
It builds out with cricoid topological structure, solves what the most common loop in mining industry industry generated by using SpanningTree
Broadcast storm problem.Experiment topology is as shown in figure 12.
Innovative technology is as follows: MOS mine Internet of Things operating system platform is by rationally and effectively using software definition
Technology for Ensuring QoS in floating resources dispatching technique and data distribution service technology is virtualized in network technology to sum up
It is a set of applied to coal information transmission field can be realized data efficient reliable transmission, Internet resources distribution according to need it is virtual
Change elastic data distributing network system.
Hardware configuration in table 4-3 physical environment
Network card configuration in table 4-4 hardware environment
The virtualization elastic data distributing network system is directed to current coal industry field and enterprise network in network layer, transport layer
Some problems existing for network system are improved.Mainly include the following:
1) IP network is realized by carrying out virtualization process to Internet resources, carrying out software function virtualization process to network element
The rationalization distribution of link circuit resource, floating resources, to maximize the service efficiency of link, improves the biography of business-critical in network
Defeated efficiency and response real-time.
2) it is realized by using the data communication normal form of publish-subscribe model and based on hard real time transmission control technology soft
Part communication construction redefines transmission data format, transmission channel, data sharing spatial cache etc., realizes one
The efficient Data distributing system of loose coupling is covered, to ensure that the service quality of data transmission, realizes critical data
The real-time transmission and distribution of flow.
Claims (10)
1. a kind of virtualization data distributes elastomeric network system, it is characterised in that: carried out to each Internet resources being arranged in the whole network
Virtualization process is established SDN network in network layer, when each Internet resources in SDN network carry out data transmission, is adopted in transport layer
With the DDS publish/subscribe mode of multiregion system.
2. virtualization data according to claim 1 distributes elastomeric network system, it is characterised in that: in SDN network system
Communication protocol use OpenFlow agreement, realized between the network equipment and SDN controller.
3. virtualization data according to claim 2 distributes elastomeric network system, it is characterised in that: based on OpenFlow's
SDN network can dispose in existing network, the hybrid network including physical network, virtual network and multiple commercial vendors equipment
Environment.
4. virtualization data according to claim 2 distributes elastomeric network system, it is characterised in that: convection current in SDN network
The dynamic high-efficiency distribution of amount is divided into after this method virtualizes the flow in network by virtualization floating resources dispatching method
General data flow and the special data flow for having QoS demand.
5. virtualization data according to claim 4 distributes elastomeric network system, it is characterised in that: virtualization floating resources
Dispatching method realizes different routing algorithms to general data flow and special data flow respectively, is allowed by link weight realization
Special data flow is avoided in the routing of general data flow.
6. virtualization data according to claim 5 distributes elastomeric network system, it is characterised in that: virtualization floating resources
Dispatching method provides special data flow the routing algorithm under more QoS index constraints, realizes all special critical data streams
Dynamic adjusts, and realizes the high-speed transfer of critical data flow；It is excessively high to there is link load in network, especially network occurs tighter
When weight packet loss phenomenon, virtualization floating resources dispatching method is using the data grouping of discarding part common data stream come Logistics networks
The QoS demand of system.
7. virtualization data according to claim 1 distributes elastomeric network system, it is characterised in that: DDS publish/subscribe mould
Formula is data-centered, and each back end (each Internet resources) transmits number in a manner of issuing or subscribe on a distribution network
According to node is publisher or subscriber, or is both publisher and subscriber.
8. virtualization data according to claim 7 distributes elastomeric network system, it is characterised in that: DDS include DCPS and
Two layers of DLRL, DCPS is a kind of data-centered information exchange model, and the entity of DCPS includes domain, domain participant, data
Writer, publisher, data reader, subscriber, theme；DLRL layers are established on the basis of DCPS, allow to service simply
It is integrated into application layer, for recombination data automatic after having data update, and all subscribers is notified to update local data.
9. virtualization data according to claim 7 distributes elastomeric network system, it is characterised in that: multiregion system is current
In system, adds new domain and have no effect to original domain.
10. virtualization data according to claim 1 distributes elastomeric network system, it is characterised in that: to virtualization data
When distribution elastomeric network system is detected, packet capturing analysis is carried out to the flow in network using Wireshark, passes through iperf
The bandwidth performance of TCP and UDP in performance testing tool test network, the performance testing tool can report the band of network flow
Width, delay jitter and data-bag lost state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910289822.9A CN109995875A (en) | 2019-04-11 | 2019-04-11 | Virtualization data distributes elastomeric network system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910289822.9A CN109995875A (en) | 2019-04-11 | 2019-04-11 | Virtualization data distributes elastomeric network system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109995875A true CN109995875A (en) | 2019-07-09 |
Family
ID=67133308
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910289822.9A Withdrawn CN109995875A (en) | 2019-04-11 | 2019-04-11 | Virtualization data distributes elastomeric network system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109995875A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111221667A (en) * | 2020-01-09 | 2020-06-02 | 浪潮软件股份有限公司 | Data distributed preprocessing method based on CMSP |
| CN111416746A (en) * | 2020-03-26 | 2020-07-14 | 西安电子科技大学 | CDN-based data distribution method |
| CN112468375A (en) * | 2021-01-26 | 2021-03-09 | 奥特酷智能科技(南京)有限公司 | Method for acquiring transmission performance evaluation data for data distribution service |
| CN114666335A (en) * | 2022-03-21 | 2022-06-24 | 北京计算机技术及应用研究所 | DDS-based distributed system load balancing device |
-
2019
- 2019-04-11 CN CN201910289822.9A patent/CN109995875A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111221667A (en) * | 2020-01-09 | 2020-06-02 | 浪潮软件股份有限公司 | Data distributed preprocessing method based on CMSP |
| CN111221667B (en) * | 2020-01-09 | 2024-03-22 | 浪潮软件股份有限公司 | Data distributed preprocessing method based on CMSP |
| CN111416746A (en) * | 2020-03-26 | 2020-07-14 | 西安电子科技大学 | CDN-based data distribution method |
| CN111416746B (en) * | 2020-03-26 | 2022-03-04 | 西安电子科技大学 | CDN-based data distribution method |
| CN112468375A (en) * | 2021-01-26 | 2021-03-09 | 奥特酷智能科技(南京)有限公司 | Method for acquiring transmission performance evaluation data for data distribution service |
| CN114666335A (en) * | 2022-03-21 | 2022-06-24 | 北京计算机技术及应用研究所 | DDS-based distributed system load balancing device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109995875A (en) | Virtualization data distributes elastomeric network system | |
| US10601656B2 (en) | Network element upgrade method and device | |
| US9760391B2 (en) | Method and apparatus for network virtualization | |
| JP6538986B2 (en) | Virtual machine packet control | |
| KR101692890B1 (en) | Chassis controllers for converting universal flows | |
| US8027354B1 (en) | Network consolidation for virtualized servers | |
| EP2774048B1 (en) | Affinity modeling in a data center network | |
| CN104009871A (en) | SDN controller implementation method and SDN controller | |
| CN103763367A (en) | Method and system for designing distributed virtual network in cloud calculating data center | |
| CN105791175A (en) | Method and equipment for controlling transmission resources in software defined network SDN | |
| CN104468358A (en) | Message forwarding method and device of distributive virtual switch system | |
| CN102195803B (en) | Data communication method and system | |
| CN113179299B (en) | Service function chain cooperative control system and method for industrial internet application | |
| CN109714238A (en) | A kind of method and apparatus for realizing inter-virtual machine communication | |
| CN103067287B (en) | Forwarding and realizing under control separation architecture the method for virtual programmable router | |
| CN113810205A (en) | Method for reporting and receiving service computing power information, server and data center gateway | |
| CN113300952B (en) | Distributed drainage system for cloud security resource pool and drainage method thereof | |
| WO2015043679A1 (en) | Moving stateful applications | |
| Aktas et al. | Scheduling and flexible control of bandwidth and in-transit services for end-to-end application workflows | |
| EP3402121A1 (en) | Method and device for policy transmission in nfv system | |
| CN104219280B (en) | A kind of intelligent use data transmission channel | |
| US20140330973A1 (en) | System and method for brokering and provisioning in high-speed networks | |
| JP5436597B2 (en) | Virtual network infrastructure control system and method | |
| CN106921532A (en) | A kind of performance management method, the device of business catenary system | |
| CN109067573A (en) | A kind of traffic scheduling method and device |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190709 |
|
| WW01 | Invention patent application withdrawn after publication |