CN108092803A - The method that network element level parallelization service function is realized in network function virtualized environment - Google Patents

The method that network element level parallelization service function is realized in network function virtualized environment Download PDF

Info

Publication number
CN108092803A
CN108092803A CN201711292739.4A CN201711292739A CN108092803A CN 108092803 A CN108092803 A CN 108092803A CN 201711292739 A CN201711292739 A CN 201711292739A CN 108092803 A CN108092803 A CN 108092803A
Authority
CN
China
Prior art keywords
service function
network element
message
function
queue
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201711292739.4A
Other languages
Chinese (zh)
Other versions
CN108092803B (en
Inventor
东方
师晓敏
罗军舟
汪立鹤
王睿
李玉萍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Jiangsu Posts and Telecommunications Planning and Designing Institute Co Ltd
Original Assignee
Southeast University
Jiangsu Posts and Telecommunications Planning and Designing Institute Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University, Jiangsu Posts and Telecommunications Planning and Designing Institute Co Ltd filed Critical Southeast University
Priority to CN201711292739.4A priority Critical patent/CN108092803B/en
Publication of CN108092803A publication Critical patent/CN108092803A/en
Application granted granted Critical
Publication of CN108092803B publication Critical patent/CN108092803B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08Configuration management of networks or network elements
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/455Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533Hypervisors; Virtual machine monitors
    • G06F9/45558Hypervisor-specific management and integration aspects
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/12Discovery or management of network topologies
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/455Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533Hypervisors; Virtual machine monitors
    • G06F9/45558Hypervisor-specific management and integration aspects
    • G06F2009/45595Network integration; Enabling network access in virtual machine instances

Abstract

The invention discloses in network function virtualized environment realize network element level parallelization service function method, including:Traverse service function chain, operation service Function Decomposition parser, each service function is decomposed into basic message process unit and is referred to as network element, is analyzed, the network element characteristic in comprehensive service function, each service function is determined to the operation of message and operation domain information and preserves.Using the operation of obtained each service function and operation domain information, first operation can parallel decision algorithm, determine can to carry out the service function combination of parallel optimization in service function chain, then perform parallel optimization algorithm, the combination of parallelization service function.With reference to can not be parallel service function and parallel optimization after service function, it is combined according to former service function sequential organization, create generation new demand servicing function chain, new demand servicing function chain effectively shortens former service function chain length, the degree of parallelism between each service function is improved, delay overhead when significantly decreasing message by service function chain.

Description

The method that network element level parallelization service function is realized in network function virtualized environment
Technical field
The present invention relates to network function virtualization (NFV) and service function energy chain (SFC) field more particularly to network functions The method that network element level parallelization service function is realized in virtualized environment.
Background technology
In the data center, multiple existing in the form of physics special equipment or virtual machine, covering the 4th is generally all deployed Layer is to the 7th layer of network function, mainly including fire wall, network address translation NAT etc..Data center flow from generate, It is transferred to the processing for terminating and being required for by a variety of service functions.And these network functions are based primarily upon special equipment reality at present Existing, this just brings poor expandability, very flexible, the update cycle is long, buy and runs the problems such as expense is big.In order to solve this A little problems, a new concept --- network function virtualize (NFV, Network Function Virtualization) because This is generated.NFV provides the new method of design, deployment and management network service to solve traditional task equipment using virtualization technology The problems brought.The main thought of NFV is the separation of the network equipment of physics and the function of running thereon.This means nets Network function (NF, Network Function) can be operated in as the example of common software Internet Service Provider (ISP, Internet Service Provider) in equipment.This allows many network functions being integrated into volume server, exchange On machine and memory, these equipment can be located in data center, in distributed network node or in end-user facilities. In this way, a given service can resolve into one group of virtual network function, then can be taken in one or more commercial physics It is achieved in the software run on business device.In addition, virtual network function (VNF, Virtualized Network Function) example can be relocated and instantiated in different network sites, new hard without requiring to buy and install Part.NFV have huge flexibility, can further to user and other service open network functions and service and faster, It is cheaper to dispose or support the ability of new network service.In order to realize these advantages, set with now traditional proprietary physics Standby realization method is compared, and NFV has these following differences:
(1) software and hardware decouples.Since network element is no longer the combination that proprietary hardware and proprietary software entity integrate, The evolution of the two is independent from each other.This allows individual development time table and soft and hardware is separately maintained.
(2) flexible network function deployment.The separation of software and hardware helps to redistribute and shares infrastructure money Source, therefore hardware and software can perform different functions in different time.This contributes to service provider in same physical New network service is quickly disposed on platform.Therefore, instantiating services group in any equipment for enabling NFV that can be in a network Part, and its connection can be set in a flexible way.
(3) dynamic scales.The software services component that network function, which is decomposed into, to be instantiated provides greater flexibility, In a manner of more dynamical, the VNF performances that the extension of more fine granularity is actual.
Based on the rise of NFV technologies, the concept of service function chain (SFC, Service Function Chain) is also bred And it gives birth to.SFC is defined and has been instantiated such a chain type, (SF, Service Function, are equivalent to orderly service function NF in NFV) set, it be responsible for traffic distribution (data plane) between SF, control and monitoring (control plane) special services/ Application program.SFC is a kind of support technology, can be flowed with flexible management special services/application, solution is provided for flow point class, And appropriate strategy is implemented with the routing of stream according to the usability status of service request and network.
Recently, VNF flexible deployments, resilient expansion are being realized using network function virtualization technology NFV and is reducing cost On the basis of, SFC also optimizes SFC network transmissions using software defined network (SDN, Software Defined Network) Flexibility and scalability.It inherently sees, is realized using NFV from traditional task equipment SF to virtual network function (VNF) Transformation provides more effective and efficient SF deployment and layout;Using SDN decoupling controls and data plane, and introduce and be suitable for Programming in SFC is abstracted, for the dynamic control to SFC topological structures and across SF flow path by.
The content of the invention
In view of the deficiencies of the prior art, the invention discloses network element level parallelization clothes are realized in network function virtualized environment The method for function of being engaged in,
Comprise the following steps:
Step 1, obtain and traverse service function chain, operation service Function Decomposition parser decompose each service function For basic message process unit, it is referred to as network element, analyzes, the network element characteristic in comprehensive each service function, determine each service function Operation to message and operation domain information simultaneously preserve;
Step 2, according to the operation of each service function obtained in step 1 and operation domain information, operation can parallelization judgement Algorithm determines that the service function combination of parallel optimization can be carried out in service function chain;
Step 3, for obtained in step 2 can Parallel Service function queue, perform parallel optimization algorithm, generate new clothes Business function, and add in new demand servicing function queue chain;For it is non-can Parallel Service function, be added into new demand servicing function queue chain In;
Step 4, with reference to the new demand servicing function that is generated in step 3 and it is non-can parallelization service function, serviced according to original Functional sequence layout is combined, and output is the new demand servicing function chain after parallel optimization.
Step 1 comprises the following steps:
Step 1-1 obtains the service function chain of administrator's input;
Step 1-2, first service function in reading service function chain, and create and be named as SFHM (Service Function HashMap, service function Hash Map) Hash Map HashMap<key,values>(unique key Key corresponds to its value values), for storing the operation of each service function key and operation domain information values, operate domain information Values includes write operation domain and read operation domain;
Step 1-3 judges next service function whether in service function chain, if it is, entering step 1-4;It is no Then enter step 1-5;
Step 1-4, operation service Function Decomposition parser:
It is network element first by current service Function Decomposition, then according to the operation of rule analysis, comprehensive each network element to message And operation domain information, determine operation and operation domain of the service function to message.For service function decomposition analysis algorithm, divide Solution preocess is mainly according to operation of each several part in service function to message, including obtaining, exporting, reading and writing (modification or add drop report Literary domain), dropping packets etc., and using these basic message process units as network element, special services work(can be realized so as to be combined into The directed acyclic graph of energy.
Analysis and synthesis:Analysis, combined process needs consider the operation and operation of each network element of composition service function Domain information is so that it is determined that the possible operation and operation domain of service function, and the operation of network element, operation domain information and each network element characteristic And user policy is related, as FW according to user sets rule and message header information (reading) network element of being classified needs basis The setting of user determines to need the message domain read.Therefore, need to obtain message network element, outgoing message to removing when analyzing, is comprehensive All network elements outside network element, discarding network element, the rule taken are:When performing read operation there are network element, which owns Seek union in the read operation domain of network element;When performing write operation there are network element, the write operation domain of all network elements of the service function is asked simultaneously Collection.By the service function decomposition analysis algorithm that above-mentioned rule is realized, using operation and the operation domain information of each network element, clothes are determined Business function is to the possible operation of message and operation domain information and is saved in SFHM;
Step 1-5 is created can Parallel Service function queue PQ (Parallel Queue, can parallel queue) and new demand servicing work( Can queue chain NC (New Chain, new demand servicing function chain), can Parallel Service function queue PQ for store can Parallel Service function Combination, new demand servicing function queue chain NC is for the new demand servicing function chain after memory parallel optimization;First in reading service function chain A service function and its operation, operation domain information, be added into can Parallel Service function queue, and labeled as team's head and tail of the queue; It is 0 that service function last-in-chain(LIC) tail tag will, which puts End, enters step 2.
Step 2 comprises the following steps:
Step 2-1, judgement can Parallel Service function queue PQ when whether being sky, if being not sky, enter step 2-2;It is no Then enter step 2-3;
Step 2-2, by the service function currently read add in can Parallel Service function queue PQ, and by team's head, tail of the queue refers to To current service function;
Step 2-3, next service function and its operation, operation domain information in reading service function chain;
Step 2-4, operation can parallel decision algorithm, to the effect that:And if only if the service function that currently reads with can All service functions in Parallel Service function queue PQ, in the case of without considering packet loss, the message behaviour of each service function Make domain without intersection or each service function according to its order (current service functional sequence is end) in PQ to same message The operation in domain (i.e. operation domain has intersection) is read after read, writes or during write after write, and the service function in PQ can be with the clothes that currently read Function of being engaged in can parallelization.Judge the service function that currently reads whether can with can be in Parallel Service function queue PQ all services Function parallelization if can be parallel, enters step 2-5;Otherwise 3 are entered step;
Step 2-5, by current service function add in can Parallel Service function queue PQ, and be marked as tail of the queue, enter Step 4;
Step 3 comprises the following steps:
Step 3-1, judge currently can Parallel Service function queue PQ squadrons head tail of the queue whether be directed toward same service function (Service Function, SF), i.e., whether only comprising a service function, if so, representing that the service function cannot be with it His adjacent service function is parallel, which is stored to new demand servicing function queue chain NC, enters step 3-2;Otherwise into Enter step 3-3;
Step 3-2, empty currently can Parallel Service function queue PQ, put team head tail of the queue for sky, enter step 4;
Step 3-3 performs parallel optimization algorithm:It first creates from network interface card and obtains message network element, realize from network interface card and obtain message Function, input terminal connects network interface card to obtain message, and output terminal initialization is set to sky;
Step 3-4 is created and is replicated distribution network element, is realized copy packet and is distributed each message copy to each Parallel Service function On function, input terminal linkup transmit message to network interface card network element output terminal, output terminal initialization be set to sky;
Step 3-5 creates message and merges network element, realizes and merges the message copy after each service function and ultimately form Outgoing message, input terminal initialization are set to sky, and output terminal initialization is set to sky;
Step 3-6, obtain currently can Parallel Service function queue PQ head of the queue service functions, and set current service function grasp Make priority P ID (Priority Identification, priority tag) numerical value as queue length, PID is for two or more The asynchronous merging priority of the same message numeric field data of message copy;
Step 3-7 deletes current service function and obtains message network element from network interface card, and it is defeated that original output network element is connected to duplication distribution Outlet;
Step 3-8 judges that service function sends whether message connects queue network element to network interface card network element input terminal, if so, 3-9 is then entered step, otherwise enters step 3-10;
Step 3-9, the queue connected by the use of step 3-8 Central Plains transmission message to network interface card network element input terminal are defeated as output queue Dequeue network element, which is replaced, sends message to network interface card network element, on the basis of the former input of queue is retained, increases and sends message to network interface card Input, enter step 3-11;
Step 3-10, it is output queue network element to replace service function and send message to network interface card network element, sends message to network interface card Network element input is the input of output queue network element;
Step 3-11 replaces dropping packets network element in service function and, to be set to abandon network element, realizes that mark dropping packets replace For the function of simple dropping packets, input is former dropping packets network element input, and output network element is output queue network element;
Step 3-12, each output queue network element output terminal network element that setting steps 3-9 and 3-10 are created are to be created in step 3-5 The message built merges network element, and connects;
Step 3-13, judge current service function whether be currently can Parallel Service function queue PQ tail of the queue, if so, 3-14 is then entered step, otherwise, enters step 3-7;
Step 3-14 creates new transmission message to network interface card network element, output terminal connection network interface card, for exporting final report Text, input terminal connection message merge network element, and so far, parallel optimization algorithm is finished, and all service functions in PQ are by simultaneously Row chemical conversion is a new service function;
Service function newly-generated in step 3-14 is added in new demand servicing function queue chain NC, performs step by step 3-15 4;
Step 4 comprises the following steps:
Step 4-1 judges service function end of chain (EOC) identifier End, be worth for 1 when, enter step 4-4;Otherwise, enter step 4-2;
Step 4-2, judges whether the service function currently read is located at former service function chain end, if so, representing former The judgement of service function chain finishes, then enters step 4-3;Otherwise, 2-1 is entered step;
Step 4-3, it is 1 to set end identifier End, represents that former service function chain judgement finishes, return to step 3;
Step 4-4 is combined according to former service function sequential organization, exports new demand servicing function queue chain NC, is parallel excellent Service function chain after change.
SFC is based primarily upon the virtual network function (VNF) of NFV technologies realization at present, and by its portion in a certain order Administration, layout recycle SDN technologies to realize the dynamic stream routing between SF.However, traditional chain type connection SFC can cause with clothes The steady-state growth of business/number of applications, the length of SFC can constantly increase, so when message passes through such SFC when institutes band The delay overhead come also can be drastically increase, SFC cannot meet stringent Heterogeneous service quality (QoS, Quality of Service) requirement and service-level agreement (SLA, Service-Level Agreement), cause the development of SFC to be absorbed in Predicament.Therefore, there are many researchs that researcher is directed to this issue-resolution at present.Main research work includes:
(1) redundant operation between network function is eliminated;
(2) it is utmostly parallel between realization network function.
In (1) is studied, often there are more redundant operation between researcher's discovery NF, such as each service function to be both needed to Message, transmission message to network interface card etc. are obtained from network interface card.Therefore, the repetitive process between NF is eliminated, NF can be optimized and serially located Manage efficiency.But although the repetitive operation reduced between NF shortens delay overhead when message passes through SFC to a certain extent, But SFC is still the serial process of use, fails to make full use of the concurrency between message operation, as SFC longer, according to So there are performance bottlenecks.
And in (2) are studied, researcher the study found that at present in data center common NF have 53.8% be can be simultaneously Capable, therefore, they propose Parallel Implementation mode NF grades a kind of.They are absorbed in single server between multiple VNF Parallel optimization, realize one for generate the NF of message copy with and a NF for being used to merge multiple message copies. Although such realization method is lower compared to delay caused by the realization method in research (1), it is needed when parallel To the redundant operation of message between the duplication operation of full message and more VNF, also by the delay overhead of introducing portion.
In conclusion the SFC for the VNF compositions realized based on NFV technologies, despite the presence of can flexible deployment, scalability it is strong The advantages that, but such SFC time delays expense it is big caused by the problems such as cannot meeting QoS and SLA be also existing, and current Research, mainly reduce and redundant operation and directly optimized using parallel mode from serial chain type, but these grind Work is studied carefully also all there are respective deficiency, still, it is contemplated that huge caused by the development trend of computer realm and parallelization Big delay reduction, parallelization mode will become following mainstream.In consideration of it, the present invention proposes one kind for network function void The method that network element level parallelization service function is realized in planization environment is divided service function by service function decomposition analysis algorithm It solves and is referred to as network element, and the network element characteristic in Analysis Service function for basic message process unit, determine each service function pair The operation of message and operation domain information, then operation can parallel decision algorithm service function is carried out can parallelization analyze, it is definite The service function combination of parallel optimization can be carried out in service function chain, and runs parallel optimization algorithm and realizes parallelization;In this base On plinth, with reference to can not be parallel service function and parallel optimization after service function, create generation new demand servicing function chain, so as to have Effect ground shortens former service function chain length, improves each service function degree of parallelism, when significantly decreasing message by service function chain Delay overhead.
Advantageous effect:
Effective part of the present invention is:
By a kind of for realizing the method for network element level parallelization service function in network function virtualized environment, so as to have Effect ground shortens former service function chain length, improves each service function degree of parallelism, significantly decreases delay overhead.
Compared with prior art, the present invention it has the following advantages:
1st, service function is resolved into network element combination by service function decomposition analysis algorithm by the present invention, more fine-grained The operation of Analysis Service function and operation domain information, whereby come determine between service function can concurrency, to be directed to service work( The concurrency analysis of energy provides reference;
2nd, carry out the parallelization of alternative service functional level using the service function parallelization of network element level, avoid between service function Redundant operation and it is across virtual machine when operating process of the network interface card to message, be service function parallelization resarch in later NFV Provide new approaches;
3rd, service function decomposition analysis algorithm, can parallelization decision algorithm and parallel optimization algorithm it is simple and effective, correctly Rate is high, and complexity is low, can be applicable in long service function chain environment;
4th, modular design is so that the degree of coupling is relatively low between the various components of system, between modules by the overall situation altogether It enjoys variable to interact, is adapted to new demand and extension.
NFV is as emerging network technology, development prospect desirable.Meanwhile the SFC implementations realized based on NFV The problem of increasingly becoming mainstream, but bringing is also urgently to be resolved hurrily.In this context, the present invention for service function chain QoS and The latency issue being primarily upon in SLA proposes to be used to realize network element level parallelization service function in network function virtualized environment, By service function decomposition analysis algorithm, resolution service function is the combination of network element, and according to special to network element in each service function Property analysis, determine each network function to the operation of message and operation domain information, rerun can parallelization decision algorithm, determine clothes The service function combination of parallel optimization can be carried out in business function chain, and parallel optimization algorithm is performed to it, generates new service work( Can, and combine can not parallelization service function, export parallel optimization after new demand servicing function chain, so as to shorten former service function Chain length reduces delay overhead.The present invention provides new approaches to solve the delay of service function chain, will be service function chain And its research and development of performance optimization is contributed share.
For the method for the present invention by service function decomposition analysis algorithm, resolution service function is the oriented nothing that network element is composed Operation and the operation domain information of wherein network element are analyzed, integrated to ring figure to determine that service function believes the operation of message and operation domain Breath, operation can parallel decision algorithm determine can Parallel Service function combine, by parallel optimization algorithm parallelization service function, and With reference to non-service function that can be parallel, the new demand servicing function chain after parallel optimization is generated, so as to effectively shorten former service work( Energy chain length, delay when substantially reducing message by service function chain.
Description of the drawings
The present invention is done with reference to the accompanying drawings and detailed description and is further illustrated, it is of the invention above-mentioned or Otherwise advantage will become apparent.
Fig. 1 service function network element level parallelizingsystem Organization Charts.
Fig. 2 service function network element level parallelization flow charts.
Fig. 3 parallel optimization algorithm flow charts.
Fig. 4 service functions network element forms example.
Fig. 5 is service function network element level parallelization example in Fig. 4.
Fig. 6 mainly realizes process schematic for service function network element level parallelizingsystem.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
The invention discloses the methods that network element level parallelization service function is realized in network function virtualized environment, need first A controller is realized on the server, for obtaining the service function chain information of administrator's input, and it is carried out parallel Change is handled, and finally exports the new demand servicing function energy chain after parallel optimization, and therefore, controller mainly includes four modules, is respectively Service function decomposition analysis module, service function can parallelization analysis module, can Parallel Service function parallelization optimization module, new clothes Business function chain generation module.Concrete methods of realizing is as follows:
In service function decomposition analysis module, it is necessary first to it obtains and travels through the service function chain of administrator's input, fortune Row service function decomposition analysis algorithm decomposes wherein each service function as basic message process unit, is referred to as network element, Ran Houfen Network element characteristic in analysis, comprehensive each service function, rule are:When there are during the network element property held read operation, all nets of the service function Seek union in the read operation domain of member;When there are during the network element property held write operation, union is sought in the write operation domain of all network elements of the service function. Each service function is determined to the operation of message according to rule and is operated domain information and is saved in HashMap<key,values> In, wherein each service function is key keys, the corresponding operation of service function and operation domain are values values.
It is as follows to perform step:
Step 1, the service function chain of administrator's input is obtained, enters step 2;
Step 2, first service function in reading service function chain, and create the HashMap for being named as SFHM<key, values>For storing the operation of each service function (key) and operation domain information (values, including write operation domain and read operation Domain), enter step 3;
Step 3, judge that next service function is then to enter step 4 whether in service function chain;Otherwise, into step Rapid 5;
Step 4, operation service Function Decomposition parser, it is basic message process unit to decompose current service function first The directed acyclic graph that referred to as network element is composed, then according to rule analysis, the comprehensive each net decomposed obtained by current service function Operation and operation domain of the member to message determine the service function to the operation of message and operation domain information and store to SFHM In, enter step 3.
Step 5, establishment can Parallel Service function queue PQ and new demand servicing function queue chain NC;In reading service function chain One service function and its operation, operation domain information, are added into PQ;It is 0 that service function last-in-chain(LIC) tail tag will, which puts End, calls clothes Function of being engaged in can parallelization analysis module.
It can be in parallelization analysis module, according to each service work(obtained in service function decomposition analysis module in service function Can operation and operation domain information, operation can parallelization decision algorithm, determine that parallel optimization can be carried out in the service function chain Service function combines.Wherein, for can parallelization decision algorithm, Main Basiss are:And if only if continuous in service function chain Service function, without considering packet loss, the message operation domain of each service function is taking without intersection or each service function according to it Order in function chain of being engaged in, can between service function to when the operation of same message domain (having intersection) is read after read or write after write Parallelization.
It is as follows to perform step:
Step 1, it is no when PQ is not empty, enter step 2;It is to enter step 3;
Step 2, the service function currently read is added in into PQ, enters step 3;
Step 3, next service function and its operation, operation domain information in reading service function chain, enter step 4;
Step 4, operation can parallel decision algorithm, judge the service function that currently reads, if can be with all clothes in PQ Business function parallelization.When can be parallel, 5 be entered step;Otherwise, calling can Parallel Service function parallelization optimization module;
Step 5, current service function is added in into PQ, calls new demand servicing function chain generation module;
Can in Parallel Service function parallelization optimization module, for service function can obtain in parallelization analysis module can Parallel optimization service function combine, input can the combination parallel, perform parallel optimization algorithm, output parallelization after new demand servicing work( Can, and be added into new demand servicing function queue chain;For it is non-can Parallel Service function, be added into new demand servicing function queue chain In.
It is as follows to perform step:
Step 1, whether judge current PQ, only comprising a service function, is, processing it is non-can Parallel Service function, into step Rapid 2;Otherwise, parallel optimization algorithm is run, enters step 3;
Step 2, represent that the service function can not be parallel, be stored to NC, and empty PQ, call the life of new demand servicing function chain Into module;
Step 3, bring into operation parallel optimization algorithm, creates new slave network interface card first and obtains message network element, realizes from network interface card The function of message is obtained, input terminal connects network interface card to obtain message, and output terminal is temporarily set to sky and enters step 4;
Step 4, create and replicate distribution network element, realize copy packet and distribute each message copy to each Parallel Service functionally Function, the new slave network interface card created in input terminal Connection Step 1 obtains message network element output terminal, and output terminal is temporarily set to Sky enters step 5;
Step 5, create message and merge network element, realize and merge the message copy after each service function and ultimately form defeated Outgoing packet, input terminal are temporarily set to sky, and output terminal is temporarily set to sky, enters step 6;
Step 6, PQ head of the queue service functions are obtained, and set current service feature operation priority P ID numerical value long for queue Degree, PID are used for the asynchronous merging priority of the same message numeric field data of multiple message copies, enter step 7;
Step 7, delete current service function and obtain message network element from network interface card, original output network element is connected to duplication distribution output End, enters step 8;
Step 8, judge that service function sends whether message connects queue network element to network interface card network element input terminal.It is then to enter Step 9;Otherwise 10 are entered step;
Step 9, replaced by the use of the queue as output queue output queue network element and send message to network interface card network element, retaining team On the basis of the former input of row, increase and send message to the input of network interface card, enter step 11;
Step 10, it is output queue network element to replace service function and send message to network interface card network element, sends message to network interface card net Member input is the input of output queue network element, enters step 11;
Step 11, replace dropping packets network element in service function and, to be set to abandon network element, realize that mark dropping packets substitute The function of simple dropping packets, input input for former abandon, and output network element enters step 12 for output queue network element;
Step 12, output queue network element output network element is set to merge network element for message, and is connected.Enter step 13;
Step 13, judge current service function whether be PQ tail of the queue, be then to enter step 14;Otherwise, 7 are entered step;
Step 14, new transmission message is created to network interface card network element, output terminal connection network interface card, to export final message, Its input terminal connects each service function output queue output terminal of above-mentioned establishment.So far, parallel optimization algorithm is finished, and realizes New service function will be generated after all service function parallelizations in PQ, enter step 15;
Step 15, NC is added into, calls new demand servicing function chain generation module;
In new demand servicing function chain generation module, with reference to can be generated after parallelization in Parallel Service function parallelization optimization module New demand servicing function and a small amount of service function for being not available for parallelization, combined according to former service function sequential organization, finally Output is the new demand servicing function chain after parallel optimization.
It is as follows to perform step:
Step 1, judge service function end of chain (EOC) identifier End, be worth for 1 when, enter step 4;Otherwise, 2 are entered step.
Step 2, judge whether the service function currently read is located at former service function chain end, if so, representing former clothes Business function chain judgement finishes, then enters step 3;Otherwise, call service function can parallelization analysis module;
Step 3, it is 1 to set end identifier End, represents that former service function chain judgement finishes, calling can Parallel Service work( It can parallel optimization module;
Step 4, combined according to former service function sequential organization, new demand servicing function queue chain is exported, as after parallel optimization Service function chain.
Embodiment
The parallelization of service function network element level is realized in view of the present invention, it is necessary to carry out network element decomposition to service function, It realizes also more convenient using decomposable, modular realization method during service function.Therefore, using current most of VNF The open source software of use be click to realize.Specifically, Click modular router are that one kind is used to build spirit The expansible software architecture of network function living and configurable, it is operated in x86 framework generic servers.One is based on The service function that click is realized is formed by the processing data packets module assembled for being known as click elements.Single click elements can be real Now simple Message processing function, if message classification, queuing, scheduling and the network equipment interact, and multiple click elements can use In combining, realizing advanced routing and bag processing function, network function is used to implement.Meanwhile user can also use click offers API add self-defined click elements.Essentially, a service function is exactly one is had by what click elements formed To acyclic figure is schemed, wherein vertex is click elements, and message is transmitted along the side of figure.Here, the present invention is only by click elements In (modification and add drop message domain) is read and write to message, the key operations such as to abandon corresponding it is of the invention defined in network element, Directed acyclic graph corresponding, that service function i.e. network element are formed.Also only consider net in implementation procedure in the present invention Service function is operated member and the influence of operation domain, and without considering the influence of other click elements, these network elements are included such as Click Elements C lassifier, it is corresponding to read network element, it realizes and carries out sort operation by user policy;IPRewriter, correspondence write net Member realizes modification IP sources, destination address;Discard, it is corresponding to abandon network element, realize simple dropping packets operation.
Therefore, in order to realize for realizing the method for network element level parallelization service function in network function virtualized environment, The service function network element level parallelizingsystem frame is as shown in Figure 1, first transport service function network element level parallel optimization controller Then row operates in service function example in the KVM virtual machines of data Layer server node on key-course server node. In virtual machine, using expanded mistake click software tool packs (realization include replicate distribution network element, message merging network element, set To abandon network element etc.) to realize service function, and using Netmap Message processings frame is accelerated to accelerate virtual machine Message processing, together When in order to ensure the performance of physical node Message processing, each data Layer server node operation Intel DPDK Message processings add Fast frame.Physical node is connected by Ethernet, and the virtual machine node in every server is used based on DPDK's Openvswitch virtual switches are attached.In key-course server node, running service function network element level is parallel Optimal controller provides the functions such as the input of service function chain, modification, deletion for administrator upwards, realizes four modules downwards: Service function decomposition analysis module, service function can parallelization analysis module, can Parallel Service function parallelization optimization module, new clothes Business function chain generation module.With reference to Fig. 2, the specific implementation step of service function network element level parallel optimization is as follows:
In service function decomposition analysis module, controller obtains the service function chain of administrator's input first, wherein often A service function is presented in a manner of click configuration files.Then service function chain is traveled through, these service functions are adopted It is formed with the click elements in click modular router software tool packs, therefore, it is necessary to perform service function and decompose to divide Algorithm is analysed, service function is decomposed into the directed acyclic graph that network element is composed by screening click elements to correspond to network element, such as Shown in Fig. 4, the characteristic of network element in each service function is then analyzed, rule is:When there are during the network element property held read operation, the service Seek union in the read operation domain of all network elements of function;When there are during the network element property held write operation, all network elements of the service function write behaviour Union is sought as domain.According to rule, so that it is determined that each network function is to the operation of message and operation domain information.
Service function decomposition analysis algorithm includes:It is network element first by current service Function Decomposition, then according to rule point Analysis, comprehensive each network element determine operation and operation domain of the service function to message to the operation of message and operation domain information.Tool For body, with reference to Fig. 3, for service function decomposition analysis algorithm, decomposable process is mainly according to each several part in service function to report The operation of text, including obtaining, exporting, reading and writing (modification or add drop message domain), dropping packets etc., and will be at these basic messages Unit is managed as network element, so as to be combined into the directed acyclic graph that can realize particular service functionality.Such as in the fire wall of Fig. 3, With reference to the message operation logic of service function, fire wall can be decomposed into from network interface card and read message (acquisition), set according to user Set pattern is then classified (reading) with message header information, abandons the message (discarding) for being unsatisfactory for rule, output meets regular report The directed acyclic graph that text is composed to four network elements such as network interface cards (output);
Analysis and synthesis:Analysis, combined process needs consider the operation and operation of each network element of composition service function Domain information is so that it is determined that the possible operation and operation domain of service function, and the operation of network element, operation domain information and each network element characteristic And user policy is related, as fire wall according to user sets rule and message header information is classified (reading) network element needs It determines to need the message domain read according to the setting of user.Therefore, need to obtain message network element, output to removing when analyzing, is comprehensive All network elements outside message network element, discarding network element, the rule taken are:When performing read operation there are network element, the service function Seek union in the read operation domain of all network elements;When performing write operation there are network element, the write operation domain of all network elements of the service function Seek union.By it is above-mentioned rule realize service function decomposition analysis algorithm, using each network element operation with operation domain information, really Service function is determined to the possible operation of message and operation domain information and is saved in SFHM;
Can be in parallelization analysis module in service function, traverse service function chain is analyzed with reference to service function operation information Obtained in module each service function operation and operation domain information SFHM, to current service function with can parallelization service function team Arrange PQ perform can parallelization decision algorithm, judge whether they can be parallel, and condition is:And if only if continuous in service function chain Service function, without considering packet loss, the message operation domain of each service function is taking without intersection or each service function according to it Order in business function chain to the operation of same message domain (having intersection) for read after read/write or write after write when, service function it Between can parallelization.As soon as so often obtain can Parallel Service function set, calling can Parallel Service function parallelization optimization module Parallelization generates a new service function, until service function chain traversal finishes.
Can the content of parallelization decision algorithm be:And if only if the service function that currently reads with can Parallel Service function team Arrange all service functions in PQ, in the case of without considering packet loss, the message operation domain of each service function without intersection or According to its order (current service functional sequence is end) in PQ, to same message domain, (i.e. operation domain has friendship to each service function Integrate) operation as read after read, write or during write after write, the service function in PQ can parallelization with the service function currently read.
Can in Parallel Service function parallelization optimization module, obtain first service function can the input of parallelization analysis module can Then Parallel Service function queue judges whether head of the queue and tail of the queue are directed toward same service function, i.e., of service function in queue Number whether only there are one, be 1 when represent the network element can not parallelization, then add in new demand servicing function queue chain;When greatly with 1, table Show that these service functions can then perform parallel optimization algorithm with parallelization, algorithm flow chart is as shown in figure 3, ultimately generate one New service function, and be added into new demand servicing function queue chain., it is necessary to judge that service function chain is after aforesaid operations No traversal finishes, and finishes, and calls new demand servicing function chain generation module, otherwise then calls service function can parallelization analysis module Read next network element.In addition, can Parallel Service function parallelization optimization module in order to realize network element level parallelization, it is also necessary to it is new The support of network element includes replicating distribution network element, message merging network element, is set to abandon network element etc., their specific work(including realization It can be as follows:
Replicate distribution network element:
Function:First, message is obtained from network interface card, message ID is stamped, to distinguish message;Then, copy packet and its message ID, and stamp default action OID (being defaulted as read operation), in addition it is also necessary to according to the operator precedence grade PID of each service function, It is accurate when message copy merges to ensure for the message copy for being sent to each service function is stamped respective priority PID Property;Finally, the exit port of corresponding message copy from each service function of correspondence is sent.
Message merges network element:
Function:First, the message copy that each inbound port receives is stored in the buffer queue that each port is monopolized, chooses and work as The preceding shortest buffer to ports queue of length obtains head of the queue message;Then, each message pair is obtained from each port queue according to message ID This, total number of copies should be equal to message and merge network element inbound port sum;Then, when being set as abandoning if there is copy operation OID, Abandon all message copies, if there is copy OID be set as read and this copy be not the last one copy when, only abandon the message Copy;Finally, determine the merging order of message (i.e. when data are different according to the modes of priority P ID from high to low according to priority When, retain the high higher data of priority), each message copy union operation is carried out, generates final message.
It is set to abandon network element:
Function:For replacing Discard elements in the click in Parallel Service function, realize that operating OID by setting is Discarding prevents message copy to be simply discarded, and increases output port, and output terminal is the output queue of service function;
The element for writing functions such as (including changing and add drop message domain) is realized in modification click, increases and operation OID is set For the function of writing.
, it is necessary to follow the service function order in former service function chain, layout group in new demand servicing function chain generation module It closes the new demand servicing function that can be generated after parallelization in Parallel Service function parallelization optimization module and is not available for parallelization on a small quantity Service function, that is, ensure utmostly to realize parallelization on the basis of service function chain function is not changed.
The main realization process of the service function network element level parallelizingsystem is as shown in fig. 6, if there are clothes shown in Fig. 4 Business function chain, then by the parallelization process, the parallelization results of Fig. 4 will be as shown in Figure 5.
It is specific real the present invention provides the method that network element level parallelization service function is realized in network function virtualized environment Now there are many method of the technical solution and approach, and the above is only the preferred embodiment of the present invention, it is noted that for this For the those of ordinary skill of technical field, without departing from the principle of the present invention, several improvement and profit can also be made Decorations, these improvements and modifications also should be regarded as protection scope of the present invention.Each component being not known in the present embodiment is available The prior art is realized.

Claims (5)

1. the method for network element level parallelization service function is realized in network function virtualized environment, which is characterized in that including following Step:
Step 1, obtain and each service function is decomposed into base by traverse service function chain, operation service Function Decomposition parser This message process unit is referred to as network element, analyzes, the network element characteristic in comprehensive each service function, determines each service function to report The operation of text and operation domain information simultaneously preserve;
Step 2, according to the operation of each service function obtained in step 1 and operation domain information, operation can parallelization decision algorithm, Determine that the service function that parallel optimization can be carried out in service function chain combines;
Step 3, for obtained in step 2 can Parallel Service function queue, perform parallel optimization algorithm, generate new service work( Can, and add in new demand servicing function queue chain;For it is non-can Parallel Service function, be added into new demand servicing function queue chain;
Step 4, with reference to the new demand servicing function that is generated in step 3 and it is non-can parallelization service function, according to former service function Sequential organization is combined, and output is the new demand servicing function chain after parallel optimization.
2. according to the method described in claim 1, it is characterized in that, step 1 comprises the following steps:
Step 1-1 obtains the service function chain of administrator's input;
Step 1-2, first service function in reading service function chain, and create the Hash Map for being named as SFHM HashMap<key,values>, for storing the operation of each service function key and operation domain information values, operate domain information Values includes write operation domain and read operation domain;
Step 1-3 judges next service function whether in service function chain, if it is, entering step 1-4;Otherwise into Enter step 1-5;
Step 1-4, operation service Function Decomposition parser, including:
It is network element by current service Function Decomposition:According to basic operation of the service function to message, including from network interface card obtain message, Read the newspaper text, write message, send message to network interface card, dropping packets, service function is decomposed into these basic message process units i.e. The directed acyclic graph that network element is composed;
Analysis and synthesis:After the directed acyclic graph for being broken down into that network element is composed in current service function, according to certain Rule analysis, comprehensive each network element are to the operation of message and operation domain information, so that it is determined that operation of the service function to message And it operates domain information and is saved in SFHM;
Step 1-5 create can Parallel Service function queue PQ and new demand servicing function queue chain NC, can Parallel Service function queue PQ For store can Parallel Service function combination, new demand servicing function queue chain NC for memory parallel optimization after new demand servicing function Chain;First service function and its operation, operation domain information in reading service function chain, being added into can Parallel Service function team Row, and labeled as team's head and tail of the queue;It is 0 that service function last-in-chain(LIC) tail tag will, which puts End, enters step 2.
3. the method according to right wants 2, which is characterized in that step 2 comprises the following steps:
Step 2-1, judgement can Parallel Service function queue PQ when whether being sky, if being not sky, enter step 2-2;Otherwise into Enter step 2-3;
Step 2-2, by the service function currently read add in can Parallel Service function queue PQ, and by team head, tail of the queue be directed toward work as Preceding service function;
Step 2-3, next service function and its operation, operation domain information in reading service function chain;
Step 2-4, operation can parallel decision algorithms:And if only if the service function that currently reads with can Parallel Service function queue All service functions in PQ, in the case of without considering packet loss, the message operation domain of each service function is without intersection or each Service function according to its order in PQ is read after read to the operation in same message domain, is write or during write after write, the service in PQ Function can with the service function that currently reads can parallelization, judge the service function currently read whether can with can Parallel Service work( All service functions in energy queue PQ are parallel, if so, entering step 2-5;Otherwise 3 are entered step;
Step 2-5, by current service function add in can Parallel Service function queue PQ, and be marked as tail of the queue, enter step 4。
4. according to the method described in claim 3, it is characterized in that, in step 3, the parallel optimization algorithm, including following step Suddenly:
Step 3-1, judge currently can Parallel Service function queue PQ squadrons head tail of the queue whether be directed toward same service function, be It is no only to include a service function, if so, representing that the service function that the service function cannot be adjacent with other is parallel, this is taken Function of being engaged in storage enters step 3-2 to new demand servicing function queue chain NC;Otherwise 3-3 is entered step;
Step 3-2, empty currently can Parallel Service function queue PQ, put team head tail of the queue for sky, enter step 4;
Step 3-3 is created from network interface card and is obtained message network element, realizes the function that message is obtained from network interface card, input terminal connection network interface card To obtain message, output terminal initialization is set to sky;
Step 3-4 is created and is replicated distribution network element, is realized copy packet and is distributed each message copy to each Parallel Service functionally Function, input terminal linkup transmit message to network interface card network element output terminal, output terminal initialization are set to sky;
Step 3-5 creates message and merges network element, realizes and merges the message copy after each service function and ultimately form output Message, input terminal initialization are set to sky, and output terminal initialization is set to sky;
Step 3-6, obtain currently can Parallel Service function queue PQ head of the queue service functions, and set current service feature operation excellent First grade PID numerical value is queue length, and it is preferential that PID is used for the asynchronous merging of the same message numeric field data of more than two message copies Grade;
Step 3-7 deletes current service function and obtains message network element from network interface card, and original output network element is connected to duplication distribution output End;
Step 3-8 judges that service function sends whether message connects queue network element to network interface card network element input terminal, if it is, into Enter step 3-9, otherwise enter step 3-10;
Step 3-9 sends message by the use of step 3-8 Central Plains and exports team to the queue that network interface card network element input terminal connects as output queue Row network element, which is replaced, sends message to network interface card network element, on the basis of the former input of queue is retained, increases and sends message to the defeated of network interface card Enter, enter step 3-11;
Step 3-10, it is output queue network element to replace service function and send message to network interface card network element, sends message to network interface card network element Input is the input of output queue network element;
Step 3-11 replaces dropping packets network element in service function and, to be set to abandon network element, realizes that mark dropping packets substitute letter The function of single dropping packets, input are former dropping packets network element input, and output network element is output queue network element;
Step 3-12, each output queue network element output terminal network element that setting steps 3-9 and 3-10 are created are what is created in step 3-5 Message merges network element, and connects;
Step 3-13, judge current service function whether be currently can Parallel Service function queue PQ tail of the queue, if it is, into Enter step 3-14, otherwise, enter step 3-7;
Step 3-14 creates new transmission message to network interface card network element, output terminal connection network interface card, for exporting final message, Input terminal connection message merges network element, and so far, parallel optimization algorithm is finished, and the service function in PQ is parallelized as one A new service function;
Service function newly-generated in step 3-14 is added in new demand servicing function queue chain NC, performs step 4 by step 3-15.
5. according to the method described in claim 4, it is characterized in that, step 4 comprises the following steps:
Step 4-1 judges service function end of chain (EOC) identifier End, be worth for 1 when, enter step 4-4;Otherwise, 4-2 is entered step;
Step 4-2, judges whether the service function currently read is located at former service function chain end, if so, representing former service Function chain judgement finishes, then enters step 4-3;Otherwise, 2-1 is entered step;
Step 4-3, it is 1 to set end identifier End, represents that former service function chain judgement finishes, return to step 3;
Step 4-4 is combined according to former service function sequential organization, new demand servicing function queue chain NC is exported, as after parallel optimization Service function chain.
CN201711292739.4A 2017-12-08 2017-12-08 Method for realizing network element level parallelization service function in network function virtualization environment Active CN108092803B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711292739.4A CN108092803B (en) 2017-12-08 2017-12-08 Method for realizing network element level parallelization service function in network function virtualization environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711292739.4A CN108092803B (en) 2017-12-08 2017-12-08 Method for realizing network element level parallelization service function in network function virtualization environment

Publications (2)

Publication Number Publication Date
CN108092803A true CN108092803A (en) 2018-05-29
CN108092803B CN108092803B (en) 2020-07-17

Family

ID=62174827

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711292739.4A Active CN108092803B (en) 2017-12-08 2017-12-08 Method for realizing network element level parallelization service function in network function virtualization environment

Country Status (1)

Country Link
CN (1) CN108092803B (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108494574A (en) * 2018-01-18 2018-09-04 清华大学 Network function parallel processing architecture in a kind of NFV
CN109842528A (en) * 2019-03-19 2019-06-04 西安交通大学 A kind of dispositions method of the service function chain based on SDN and NFV
CN110022230A (en) * 2019-03-14 2019-07-16 北京邮电大学 The parallel dispositions method of service chaining and device based on deeply study
CN110227265A (en) * 2019-06-18 2019-09-13 贵阳动视云科技有限公司 A kind of computer graphical resource share method and device
CN110557343A (en) * 2018-05-31 2019-12-10 中国电信股份有限公司 SFC service data forwarding method and SFC network system
US10805164B2 (en) 2018-12-14 2020-10-13 At&T Intellectual Property I, L.P. Controlling parallel data processing for service function chains
CN112333035A (en) * 2020-12-30 2021-02-05 中国人民解放军国防科技大学 Real-time hybrid service function chain embedding cost optimization method and device
CN112997455A (en) * 2018-11-14 2021-06-18 中兴通讯股份有限公司 Communication method for service framework
CN113411207A (en) * 2021-05-28 2021-09-17 中国人民解放军战略支援部队信息工程大学 Service function circulation arrangement basic platform and method of intelligent network service function chain
US11146506B2 (en) 2018-12-14 2021-10-12 At&T Intellectual Property I, L.P. Parallel data processing for service function chains spanning multiple servers
CN114124713A (en) * 2022-01-26 2022-03-01 北京航空航天大学 Service function chain arrangement method for operation level function parallel and self-adaptive resource allocation
US11979290B2 (en) 2022-11-14 2024-05-07 At&T Intellectual Property I, L.P. Controlling parallel data processing for service function chains

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104780099A (en) * 2014-01-10 2015-07-15 瞻博网络公司 Dynamic end-to-end network path setup across multiple network layers with network service chaining
CN105406992A (en) * 2015-10-28 2016-03-16 浙江工商大学 Business requirement transformation and deployment method for SDN (Software Defined Network)
CN105721535A (en) * 2014-12-23 2016-06-29 英特尔公司 Parallel processing of service functions in service function chains

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104780099A (en) * 2014-01-10 2015-07-15 瞻博网络公司 Dynamic end-to-end network path setup across multiple network layers with network service chaining
CN105721535A (en) * 2014-12-23 2016-06-29 英特尔公司 Parallel processing of service functions in service function chains
CN105406992A (en) * 2015-10-28 2016-03-16 浙江工商大学 Business requirement transformation and deployment method for SDN (Software Defined Network)

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
FANG DONG: "Accelerating skycube computation with partial and parallel processing for service selection", 《IEEE》 *
韩青: "一种保证NFV 可靠性的最优备份拓扑生成方法", 《计算机应用研究》 *

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108494574A (en) * 2018-01-18 2018-09-04 清华大学 Network function parallel processing architecture in a kind of NFV
CN108494574B (en) * 2018-01-18 2020-06-19 清华大学 Network function parallel processing infrastructure in NFV
CN110557343A (en) * 2018-05-31 2019-12-10 中国电信股份有限公司 SFC service data forwarding method and SFC network system
CN112997455A (en) * 2018-11-14 2021-06-18 中兴通讯股份有限公司 Communication method for service framework
US11146506B2 (en) 2018-12-14 2021-10-12 At&T Intellectual Property I, L.P. Parallel data processing for service function chains spanning multiple servers
US11502910B2 (en) 2018-12-14 2022-11-15 At&T Intellectual Property I, L.P. Controlling parallel data processing for service function chains
US10805164B2 (en) 2018-12-14 2020-10-13 At&T Intellectual Property I, L.P. Controlling parallel data processing for service function chains
US11962514B2 (en) 2018-12-14 2024-04-16 At&T Intellectual Property I, L.P Parallel data processing for service function chains spanning multiple servers
US11582167B2 (en) 2018-12-14 2023-02-14 At&T Intellectual Property I, L.P. Parallel data processing for service function chains spanning multiple servers
CN110022230A (en) * 2019-03-14 2019-07-16 北京邮电大学 The parallel dispositions method of service chaining and device based on deeply study
CN109842528B (en) * 2019-03-19 2020-10-27 西安交通大学 Service function chain deployment method based on SDN and NFV
CN109842528A (en) * 2019-03-19 2019-06-04 西安交通大学 A kind of dispositions method of the service function chain based on SDN and NFV
CN110227265B (en) * 2019-06-18 2020-06-26 贵阳动视云科技有限公司 Computer graphic resource sharing method and device
CN110227265A (en) * 2019-06-18 2019-09-13 贵阳动视云科技有限公司 A kind of computer graphical resource share method and device
CN112333035A (en) * 2020-12-30 2021-02-05 中国人民解放军国防科技大学 Real-time hybrid service function chain embedding cost optimization method and device
CN113411207A (en) * 2021-05-28 2021-09-17 中国人民解放军战略支援部队信息工程大学 Service function circulation arrangement basic platform and method of intelligent network service function chain
CN113411207B (en) * 2021-05-28 2022-09-20 中国人民解放军战略支援部队信息工程大学 Service function circulation arrangement basic platform and method of intelligent network service function chain
CN114124713A (en) * 2022-01-26 2022-03-01 北京航空航天大学 Service function chain arrangement method for operation level function parallel and self-adaptive resource allocation
US11979290B2 (en) 2022-11-14 2024-05-07 At&T Intellectual Property I, L.P. Controlling parallel data processing for service function chains

Also Published As

Publication number Publication date
CN108092803B (en) 2020-07-17

Similar Documents

Publication Publication Date Title
CN108092803A (en) The method that network element level parallelization service function is realized in network function virtualized environment
Karlin et al. VERA: An extensible router architecture
Zhang et al. HyperVDP: High-performance virtualization of the programmable data plane
CN102334112B (en) Method and system for virtual machine networking
KR101827369B1 (en) Apparatus and method for managing data stream distributed parallel processing service
CN105229973A (en) The formation of network and the visualization device of action
CN104052685B (en) Flow table update method and flow table updating device
WO2015147913A1 (en) System and methods for collaborative query processing for large scale data processing with software defined networking
CN101242362B (en) Find key value generation device and method
CN105308920B (en) Multicast Flows reorder scheme
Zhao et al. A unified modeling framework for distributed resource allocation of general fork and join processing networks
Moro et al. A framework for network function decomposition and deployment
JP5842174B2 (en) Bus control device, relay device and bus system
Rygielski et al. Modeling and prediction of software-defined networks performance using queueing petri nets
RU2584471C1 (en) DEVICE FOR RECEIVING AND TRANSMITTING DATA WITH THE POSSIBILITY OF INTERACTION WITH OpenFlow CONTROLLER
CN107529352A (en) Programmable switch (PIPS) for the agreement independence of the data center network of software definition
Li et al. A multi-stage approach for virtual network function migration and service function chain reconfiguration in NFV-enabled networks
Taher Testing of floodlight controller with mininet in sdn topology
CN109213566A (en) Virtual machine migration method, device and equipment
Cao et al. A study on application-towards bandwidth guarantee based on SDN
Lixin et al. Software-Defined Protocol Independent Parser based on FPGA
Głąbowski et al. Modelling of single-sided multiservice switching structures
Nichols Simulation and analysis of adaptive routing and flow control in wide area communication networks
CN107171953B (en) Virtual router implementation method
Robles-Gomez et al. Evaluation of a Fabric Management Mechanism for Advanced Switching in Presence of Traffic

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
CB02 Change of applicant information
CB02 Change of applicant information

Address after: 210019 No. 58 East Street, Nanxi River, Jianye District, Nanjing, Jiangsu

Applicant after: Zhong Tong clothing consulting and Design Research Institute Co., Ltd.

Applicant after: Southeast University

Address before: 210019 No. 58 East Street, Nanxi River, Jianye District, Nanjing, Jiangsu

Applicant before: Jiangsu Posts & Telecommunications Planning and Designing Institute Co., Ltd.

Applicant before: Southeast University

SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant