CN109714275B

CN109714275B - SDN controller for access service transmission and control method thereof

Info

Publication number: CN109714275B
Application number: CN201910006233.5A
Authority: CN
Inventors: 张崇富; 林晨; 邱昆
Original assignee: University of Electronic Science and Technology of China
Current assignee: University of Electronic Science and Technology of China
Priority date: 2019-01-04
Filing date: 2019-01-04
Publication date: 2022-03-15
Anticipated expiration: 2039-01-04
Also published as: CN109714275A

Abstract

The invention discloses an SDN controller for accessing service transmission and a control method thereof, wherein the SDN control comprises an SDN controller interface, a network topology management module, a route management module, a flow management module, a route calculation module, a call forbidding module, a transmission traffic control module, an equipment management module, a network measurement module, a QOS module and an SDN standard control module; and SDN controller interfaces include controller-to-repeater interfaces, controller-to-controller interfaces, and controller-to-service interfaces. Meanwhile, the invention also discloses a dynamic differentiation control method based on the SDN controller, which is used for routing and forwarding the service flow to be forwarded by distinguishing the shunting mode of the large and small flows and based on a two-stage queue and a dynamic routing mechanism of the SDN. The implementation of the invention can ensure the quality of the end-to-end network service flow in the network with the scale of the metropolitan area and above, namely ensure the bandwidth of the high-priority service flow.

Description

SDN controller for access service transmission and control method thereof

Technical Field

The invention belongs to the field of communication technology, and particularly relates to a centralized control method adopting a software defined network structure.

Background

The Quality of Service (QoS) refers to a Quality of Service capability of an IP network, that is, a Service required by a specific Service is provided to an IP network spanning multiple underlying network technologies (FR, ATM, Ethernet, SDH, etc.). Technical indexes for measuring the IP QoS comprise: bandwidth/throughput, latency, jitter, packet loss rate, and availability, where bandwidth/throughput refers to the average rate of a particular application traffic flow between two nodes of a network; latency refers to the average round trip time for a data packet to travel between two nodes of a network; jitter refers to the variation in time delay; the packet loss rate is the percentage of lost messages in the network transmission process and is used for measuring the capability of correctly forwarding user data by the network; availability refers to the percentage of time that the network can provide service to the user.

Different services have different requirements on IP QoS technical indexes, and by effectively implementing various IP QoS technologies, network managers can effectively control network resources and use thereof, and can better integrate various services such as voice, video, data and the like on a single IP network platform.

The network architecture of the Internet is based on end-to-end parameters, where the network supports minimization and the end-hosts are responsible for most communication tasks. This best effort network architecture is adaptable when the network requirements are primarily reliability. However, in multimedia traffic transmission, the requirement for timely delivery is superior to reliability. Multimedia streaming applications have stringent delay requirements that cannot be guaranteed in a best effort network transport architecture. Therefore, there is a need for a network to provide a method for multimedia streaming that can guarantee QoS (quality of service guarantee). To this end, several QoS architectures have been proposed by the IETF, but none are fully successful and implemented for global deployment.

The integrated QoS architecture and the differentiated services QoS architecture are all completely distributed hop-by-hop routing architectures based on the current internet, and lack of regulation and control capability on resources of the whole network. Even though MPLS (multi-label protocol switching) provides a partial solution through its ultra-fast switching capability, it still lacks real-time reconfigurability and network adaptability.

As (autonomous system) system refers to a group of routers and networks under the control of one authority. It can be a router directly connected to a LAN and also connected to Internet; it may be a plurality of local area networks interconnected by an enterprise backbone network. All routers in an autonomous system must be interconnected, run the same routing protocol, and assign the same autonomous system number. But one AS can only run one routing protocol.

An SDN (Software defined Network-Define Network) is a novel Network architecture, and the design concept of the SDN is to separate a control plane and a data forwarding plane of a Network, so that programmable control of bottom-layer hardware is realized through a Software platform in a centralized controller, and flexible Network resource allocation as required is realized. In the SDN network, the network device is only responsible for pure data forwarding, and may adopt general hardware; the original operating system responsible for control is abstracted into an independent network operating system which is responsible for adapting to different service characteristics, and the communication among the network operating system, the service characteristics and the hardware equipment can be realized through programming.

The traditional QoS mechanism is designed for the Internet, is established on a completely distributed hop-by-hop routing type system structure of the Internet, lacks a uniform global view of the whole network resource distribution, and is difficult to popularize and apply. The SDN has the characteristic of centralized control, the QoS strategy can be easily issued through the centralized controller, centralized management control over all network devices and the flow of the whole network is achieved, flexible QoS service strategy selection can be completed, and the consistency of the QoS strategy can be guaranteed. Current SDN network development is still deficient in research into QoS services. As with conventional networks, the network will typically treat all data streams equally for traffic of users of different priorities or different types of services. When the network bandwidth condition cannot meet the requirement, network congestion is generated, and at this time, traffic with higher priority, such as a conference, a video or data of a high-priority user, cannot pass preferentially, so that the user payment model and the service quality model are not equal.

Disclosure of Invention

The invention aims to: in view of the above problems, an SDN controller capable of effectively guaranteeing transmission quality and transmission rate is provided.

The invention discloses an SDN controller for accessing service transmission, which comprises an SDN controller interface, a network topology management module, a route management module, a flow management module, a route calculation module, a call forbidding module, a transmission traffic control module, an equipment management module, a network measurement module, a QOS module and an SDN standard control module;

wherein the SDN controller interface comprises:

a controller-forwarder interface for the SDN controller to provide a secure channel for the forwarder;

a controller-controller interface for information interaction between SDN controllers;

a controller-service interface for a service provider to set management rules;

the network topology management module is used for acquiring current network topology information based on the network data received from each transponder and sending the current network topology information to the route management module and the route calculation module;

the route management module: the router working information of each router in the current network is determined;

a flow management module for efficient flow management based on obtaining flow management definitions through a controller-service interface and through link aggregation;

the route calculation module is used for calculating and determining route flows of different service types;

a call barring module: when the requested QoS parameters cannot meet the requirements, the module rejects/blocks the current request and informs the SDN standard control module to execute a corresponding instruction;

a transmission traffic control module: the QoS system is used for determining whether the data flow is consistent with the QoS request parameter and executing a preset regulation rule when the data flow is inconsistent with the QoS request parameter;

a device management module: the system is used for recording and discovering the network equipment in use, tracking the transfer of different equipment in the network and recording the configuration information of the different equipment;

a network measurement module: the system comprises a route calculation module, a link utilization rate acquisition module, a data packet loss rate acquisition module and a data transmission module, wherein the route calculation module is used for acquiring current network information, including the link utilization rate, the data packet loss rate and network time delay, and transmitting the current network information to the route calculation module in real time;

a QoS module: acquiring definitions of different data flows of a user, and distinguishing QoS service requirements of the data flows through the definitions; the SDN standard control module is used for generating a flow table, establishing safety information and issuing the flow table;

an SDN standard control module: for managing network state, controlling sessions of the controller with forwarding layer devices, and controlling network traffic.

Meanwhile, the invention also discloses a dynamic differentiation control method based on the SDN controller, which comprises the following steps:

step 1: QOS flow pre-processing in the network:

setting a type identifier of a multimedia service flow, and defining a data service flow in the multimedia service flow as a small flow and a video service flow as a large flow; setting different priorities for each type simultaneously;

step 2: based on two-stage queues and a dynamic routing mechanism of the SDN, the service flow to be forwarded is routed and forwarded:

each output port of the SDN switch is configured to maintain two forwarding queues with different priorities: a high priority queue and a low priority queue; and the queue scheduler of the output port adopts an absolute priority scheduling mode: only when the high priority queue is empty, sending the packets in the low priority queue;

storing the small current flow into a high priority queue and storing the large current flow into a low priority queue;

when the output port is judged to be idle (not occupied), if the output port is idle, an absolute priority scheduling mode is adopted; otherwise, selecting the path with the lightest load for the small flow in the high-priority queue to forward until the output port is idle;

meanwhile, when routing forwarding is carried out, the path selection mode is as follows:

constructing a network topology structure diagram based on the network resource information, wherein the expression mode of the network topology structure diagram is an acyclic directed graph (DAG diagram);

based on the network topology structure chart, searching the first K shortest paths in all paths from the source switch to the destination switch;

and respectively counting the flow quantity in the forwarding queue of each hop switch port in each of the K paths (namely, counting both the high-priority queue and the low-priority queue) based on the network resource information, accumulating the flow quantity in the forwarding queues of all hops to obtain the total flow quantity of each path, and taking the path with the minimum total flow quantity in the K paths as an optimal path.

In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:

the method can ensure the quality of the end-to-end network service flow in the network with the size of a metropolitan area and above, namely ensure the bandwidth of the high-priority service flow.

Drawings

FIG. 1 is a schematic diagram of the SND controller of the present invention in an exemplary embodiment;

fig. 2 is a diagram illustrating a part of function module call interaction process of a QOS module according to an embodiment;

FIG. 3 is a diagram illustrating a detailed operation flow of a QoS module according to an embodiment;

fig. 4 is a network topology diagram of an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.

In a data center network, most of the traffic forms are similar, i.e., traffic flows in which large flows and small flows are mixed. The small stream is sensitive to time delay but has low requirement on bandwidth, and the large stream has high requirement on bandwidth but is not sensitive to time delay. In the present embodiment, a multimedia stream is treated as a big stream in a data center network, and a data stream is treated as a small stream.

The forwarding and control separation characteristics of the SDN can effectively realize gradual integration of the equipment, and reduce the hardware cost of the equipment. The centralized control logic characteristic of the SDN can gradually realize centralized management and global optimization of the network, effectively improve the operation efficiency and provide end-to-end network service; the network capacity virtualization and the network capacity openness of the SDN are beneficial to the intelligent and open development of telecommunication operator networks, the development of richer network services and the increase of income. Therefore, the SDN controller with the following structure is arranged for realizing the processing of the large and small flows.

The SDN controller comprises an SDN controller interface, a network topology management module, a route management module, a flow management module, a route calculation module, a call forbidding module, a transmission traffic control module, an equipment management module, a network measurement module, a QOS module and an SDN standard control module;

referring to fig. 1, the SDN controller interface includes:

(1) and the controller-repeater interface is used for providing a secure channel for the repeater by the SDN controller so as to use an OpenFlow protocol for information interaction. The controller is responsible for sending flow tables associated with the data flows, mapping the network-wide topology information of the network based on the network state information of the individual repeaters, and simultaneously detecting the network.

(2) And the controller-controller interface is used for information interaction between the controller and the controller, so that the controller and the controller share necessary information to cooperatively manage the whole network. Since a single controller architecture cannot be well expanded when a network is large, and the number of OpenFlow nodes increases, a multi-controller architecture is necessary.

(3) A controller-service interface for a service provider to set management rules including flow definitions of data packets, routing rules, control instructions to the SDN controller, and the like. I.e. the controller provides an open and secure interface via which the service provider can set up administrative rules, such as flow definitions for new data packets or even new routing rules for these data packets. The interface also provides a real-time interface to instruct the controller when a new application starts a data stream.

The network topology management module is used for acquiring current network topology information (namely discovering and maintaining network connection) based on the network data received from each transponder and sending the current network topology information to the route management module and the route calculation module;

the route management module: for determining router operational information (including router availability and packet forwarding performance) for each router in the current network to facilitate route computation;

and the route calculation module (path planning module) is used for calculating and determining the route flows of different service types. Various routing algorithms can be built in the module, and different routing algorithms run in parallel to meet the performance requirements and targets of different streams. And the network topology and the route management information are processed by the route calculation module along with the service reserved resources, and forwarding and bandwidth allocation rules are calculated for different types of data streams. The path planning module realizes the calculation of the path by exchanging topology and network state information provided by the path planning module and the network topology management module, the equipment module and the network measurement module;

a call barring module: when the requested QoS parameters cannot meet the requirements (i.e. there is no feasible route), this module rejects/blocks the current request and informs the SDN standard control module to execute the corresponding instructions;

a transmission traffic control module: the method is used for determining whether the data flow is consistent with the QoS request parameter and executing a preset regulation rule (such as selective packet loss) when the data flow is inconsistent with the QoS request parameter;

a network measurement module: the routing computation module is used for obtaining current network information including link utilization rate, data packet loss rate and network time delay and transmitting the current network information to the routing computation module in real time.

A QoS module: acquiring definitions of different data flows of a user, and distinguishing QoS service requirements of the data flows through the definitions; and the system is used for describing and analyzing the routing flows of different service types generated by the routing calculation module by using a DAG (directed acyclic graph model) model, mapping the routing flows into an SDN controller command, calling an SDN standard control module to generate a flow table, establishing safety information and issuing the flow table. The QoS module of the invention comprises two sub-modules which are respectively a QoS strategy configuration module and a strategy deployment module, wherein the QoS strategy configuration module is used for acquiring the definitions of different data streams of a user and distinguishing the QoS service requirements of the data streams through the definitions; the strategy deployment module is used for describing and analyzing the routing flows of different service types generated by the routing calculation module by using a DAG (directed acyclic graph model) model, mapping the routing flows into an SDN controller command, calling an SDN standard control module to generate a flow table, establishing safety information and issuing the flow table.

Referring to fig. 2, part of the function module call interaction process of the QOS module of the present invention is:

the user inputs the service requirement by calling the QoS module, and the network measurement module helps the equipment management and network topology management module to maintain the resource data of the current network. The path planning module is the core, and determines the routing forwarding and bandwidth allocation strategies according to the resource data of the network topology management and equipment management module and the strategy scheme in the QoS module, so that the QoS module packs the flow table and sends the flow table to the switch.

Referring to fig. 3, the specific working flow of the QoS module of the present invention is:

step 1: loading a QoS module;

step 2: reading the QoS service state, and if the current service state is a QoS configuration waiting request, executing the step 3; if the current service state is waiting for network connection, executing step 9;

and step 3: monitoring whether a configuration request exists or not, if so, analyzing the request and then executing a step 4;

and 4, step 4: judging whether the QoS is opened or not, if so, executing a step 5; otherwise, executing step 6;

and 5: opening QoS service;

step 6: judging whether a QoS deployment request exists or not, if so, executing a step 7; otherwise, executing step 8;

and 7: processing the topology change rule, storing the strategy analyzed from the configuration request into the controller, and writing the flow table into the switch;

and 8: storing the strategy analyzed from the configuration request into the controller, setting the QoS service state as waiting for the QoS configuration request, and jumping to the step 3;

and step 9: monitoring whether a network connection request exists, if so, executing a step 10; otherwise, setting the QoS service state as waiting for network connection, and continuing to execute the step 9;

step 10: after reading the network configuration information, starting QoS service; and setting the QoS service state as waiting for the QoS configuration request, and jumping to the step 3.

The SDN controller provided by the invention realizes the control process of dynamic differentiation of access service transmission, and comprises the following steps:

(1) QOS flow pre-processing in a network.

When a packet arrives at the router, it examines the packet's source and destination addresses and routing table entries and forwards the packet according to predetermined rules, as configured by the network operator. OpenFlow, on the other hand, provides a method for flexibly defining different flows, enabling different flows to be associated with a set of actions and rules. For example, the same type of flow may be forwarded using an optimal routing algorithm, while other flows may follow a manually configured route. Thus, each flow (i.e., packet) of the network layer may be treated differently.

In an open QOS, a flow may be defined in a variety of ways. The same stream may contain packets of the same or different types. For example, a packet with TCP port number 80 (bit HTTP reserved) may be a flow definition, or a packet with RTP header may be a flow definition, representing a flow that carries voice, video, or both. In essence, the stream may be set as a combination of header fields. But the network operator should also consider the limited processing power of the network equipment. To avoid lookup of complex flow tables as much as possible, the flow definitions should be smartly set and aggregated as much as possible. The open QoS framework makes use of OpenFlow flow-based forwarding features to enable the controller to distinguish between data and multimedia flows.

The multimedia stream may be defined using the following packet header fields or values:

traffic class header field of MPLS (MPLS in MPLS);

a type of service field (ToS (type of service) of IPv4 of IPv 4);

a Traffic class field in IPv6 (Traffic class field in IPv 6);

if the multimedia server is known, adopting a source IP address mode;

a transmission source address or a destination port number.

Defining a stream from packet header information in the lower layers (data link layer and network layer in a TCP/IP five-layer network architecture) is desirable because the lower layer packet header information is less complex than the upper layer (transport layer) processed header information.

Thus, since the MPLS field is typically considered information between the data link layer and the network layer (L2.3) and can provide the capability of ultra-fast switching, the present invention employs the MPLS field to define multimedia flows.

However, in some special cases, for better packet type differentiation, it may also be necessary to define the flow with upper layer header information, and OpenFlow supports defining the flow with a header. Furthermore, the flow definition may not be dependent on the current IP, and any addressing scheme with service level information may be used to define the multimedia type flow.

(2) SDN based two-level queuing and dynamic routing mechanisms.

And (2-1) adopting two-stage queue management in a QoS mechanism of the SDN, distinguishing to treat the small flows and the small flows, putting the small flows (data flows) into a high-priority queue, and putting the large flows (video service flows) into a low-priority queue, thereby ensuring the time delay of the small flows and the bandwidth of the large flows.

Each output port of the SDN switch maintains two queues of different priorities. The queue scheduler uses an absolute priority scheduling method, i.e. only when the high priority queue is empty, packets in the low priority queue are allowed to be sent out. When a packet is in the process of being transmitted and a high priority packet arrives, there are two processing methods: one is to allow the end of transmission of the packet currently being transmitted before transmitting a high priority packet, called non-preemptive priority; the other is that the high priority packet interrupts the transmission of the current low priority packet, and then transmits the low priority packet after the transmission of the high priority packet is finished, which is called as a high priority.

If the output port transmission is considered to be of a non-preemptive nature, then newly arriving packets (even high priority streamlet packets) have to wait when there are packets being transmitted, resulting in additional latency.

The main reasons that the two-stage queue management mechanism can improve the delay performance of the streamlets are as follows: the high priority setting improves the efficiency of transmitting streamlets. If two-stage queues are not used, large flows will cause damage to small flows.

However, since it is necessary to wait for the end of transmission of the packet being sent, the delay experienced by the streamlets not only includes queuing delay, but also adds extra waiting delay, and the delay gain caused by the two-stage queue management mechanism is reduced. This is because two levels of queue management cannot eradicate the additional added latency.

It can be seen that when the output port processing power is considered to be non-aggressive, the streamlets will experience additional latency which the two-stage queue management cannot overcome. The invention provides a new mechanism combining the hierarchical queue management and the dynamic routing to ensure the time delay of small flow and the bandwidth of large flow under the condition that an output port is not in a preemption characteristic.

The two-stage queue management mechanism can reduce the queuing delay of the small flow packets. In order to reduce waiting time delay, when a link is busy, a new route is selected for small flow packets to forward.

When the output port is not occupied, the flow is distinguished, the small flow enters a high priority queue, the large flow enters a low priority queue, and the packet is scheduled and sent by adopting absolute priority. When the output port is occupied, after two-stage enqueue management is completed, a path with the lightest load is selected for the small flow by using a dynamic routing module (namely a scheduling routing selection module) for forwarding so as to ensure time delay.

In routing, the invention can use the number of flows processed on the port in a time window and entering different queues to measure the load. The reason for choosing the number of flows is to better interface with the two previous levels of queue management. The advantage of selecting the number of streams as the measurement standard is mainly that the number of streams can be respectively counted for large streams and small streams, so as to achieve the purpose of distinguishing the loads generated by the large streams and the small streams and better avoid the time delay influence generated by the large streams on the small streams.

The SDN has programmable flexible scheduling of traffic and the property that it can collect network-wide information is such that: different routing rules are defined for different data flows. Controllers in an SDN network are the brains in the network that determine routing changes, different algorithms in the controllers associated with different data flows may produce different routing choices, and the controllers tell the network forwarders the rules of how to direct traffic flows.

The two-stage queue scheduling mechanism in the output port of the invention is formed by two modules of two-stage queue management and dynamic routing scheduling. The two-stage queue management is divided into two stages of data collection and issuing queue decision. The OpenFlow switch is responsible for collecting network flow information and reporting the flow information to the controller. The control program in the controller, which is responsible for queue management, uses the collected information to distinguish the multimedia stream from the data stream (i.e., the big stream and the small stream), and issues an enqueue policy for each stream.

The dynamic routing scheduling is divided into two stages of data collection and decision issuing.

First, data collection is performed. The controller collects network traffic through openflow protocol, collects network topology information (calls the network topology management module of the SDN controller of the present invention) through DDLP protocol, and provides the information to the routing decision application.

Second, a routing decision is made (call routing computation module). And the routing decision application program utilizes the acquired topological information and the network flow, and then selects an optimal route based on a dynamic routing algorithm of the number of the flows, thereby finally realizing the purpose of ensuring the small flow delay and the large flow bandwidth.

(2-2) routing in the SDN network by using the SDN controller, different from the traditional network, due to the characteristic of centralized control of the SDN network, the controller can make an optimal routing decision according to the overall situation of network information without using a distributed routing algorithm of switching nodes as in the traditional network. In a network environment, it is necessary to know real-time dynamic network information, which mainly includes dynamically changing network resources and network traffic conditions. The network resource information mainly includes switch, port and host information, and the network traffic information mainly includes flow-based traffic statistics and port-based traffic statistics information. Based on the currently acquired network resource information and network traffic information, the controller can make the most correct routing decision according to the current information, thereby realizing the communication of the network.

Based on this, global network information needs to be obtained first, and then routing is made. The whole dynamic routing mechanism comprises three parts: the system comprises a network resource sensing part, a network flow monitoring part and a path selection part. The network resource sensing part is used for detecting network topology information, the network flow monitoring part is used for detecting network flow information, the two parts transmit the information to the path selection part, and the path selection part is responsible for making correct routing by using the obtained information. The specific implementation processes of the network resource sensing part, the network flow monitoring part and the path selection part are as follows:

1) a network resource aware portion.

The network resource sensing module is used for detecting real-time changes of network resources, including topology information and host information. Centralized control of the SDN network allows the controller to make optimal decisions based on global information without the need to use distributed routing algorithms on the switching nodes.

2) And a network flow monitoring part.

The information of the network includes information such as a logical link in addition to the physical resource information. In addition, the statistical condition of the data traffic of the network is acquired, so that the method plays an important role in preventing network faults, reasonably optimizing the network and the like. The network flow monitoring module realizes monitoring of port flow and flow table item flow. So that the application can periodically acquire the traffic information.

In addition to physical resource information, the information of the network also includes information such as logical links, data flow traffic statistics, etc. In addition, statistics of network data traffic play an important role in preventing network failures and in reasonably optimizing the network. The network flow monitoring module monitors the flow of the flow table entry and the flow of the port. So that the application program can circularly acquire the flow information.

3) A path selection section.

The path selection part is used for selecting paths based on the traffic information provided by the network traffic monitoring module and the network resource information provided by the network resource sensing module.

The invention carries out optimal path selection based on the number of streams of a port queue, and the realization steps are as follows:

a. finding out the first K shortest paths from the source switch to the destination switch, wherein K is an empirical preset value;

b. counting the number of flows in a corresponding queue of each hop switch port of the K paths, counting the accumulated flow number of all hops of each path of the K paths, and taking the path with the minimum accumulated flow number as an optimal path;

in step a, the first K shortest paths in all paths from the source switch to the destination switch need to be found. The perception of the network topology resources is realized through a network resource perception module, and the shortest path is calculated. Firstly, the controller acquires network link information by issuing an LLDP message, and then generates a network topology map by using the network information. The network aware application stores the topology information in a DAG graph and uses a function to find the shortest path from the source switch to the destination switch.

In step b, the accumulated flow number of each hop path of the K paths is compared. The measured parameter is the queue length of the next hop routing port, where the queue length is replaced by a coarse number of queues.

In the step a, when the network topology graph is generated, the corresponding network topology graph is generated based on the expression (policy deployment module internal mechanism) of the SDN network topology structure of the DAG. The DAG graph is an acyclic directed graph and is widely used for describing business logic. Because the DAG can intuitively express the flow direction and the node relation of the data, the SDN network topology structure is modeled by using the DAG.

Examples

Referring to the network architecture shown in fig. 4, there will be two different transmission flows based on whether QoS (quality of service assurance) service registration servers are used.

1. The transmission flow of the QoS (quality of service assurance) service registration server is not adopted:

(1-1) in the network topology shown in fig. 4, two devices connected by solid lines are intercommunicated on the network, and the SDN controller is connected to other devices by dotted lines, representing a secure tunnel, which is also intercommunicated on the network.

The scope controlled by the SDN controller is an Autonomous System (AS), and the router S1 is defined AS an edge router of the AS. Each of autonomous network 1(AS 1) and autonomous network 2(AS 2) has its corresponding SDN controller, which is not shown in the figure. S denotes an SDN switch in the SDN network.

(1-2) the multimedia service subscriber h1 located at AS 1 transmits a request packet to the multimedia server h2 located at AS 3 to acquire a multimedia service.

(1-3) when the request packet sent by h1 enters the SDN network, the service request packet enters the SDN network boundary network device, and a packet _ in packet is generated at S1.

(1-4) packet _ in data packet arrives at the SDN controller located in the AS 2, obviously, packet _ in is a service request data packet, and AS a data flow, the SDN controller does not call a relevant QoS module, the SDN controller sends broadcast information to all network devices located in the AS 2, and all SDN switches receive relevant signaling and establish corresponding security channels (pipelines) with the SDN (i.e., dotted lines in the figure).

(1-5) after the security channel is established, the network device in the AS 2 network sends information of the network device and surrounding neighbors to the SDN controller, so that the SDN controller obtains network information (flow information, bandwidth information, congestion information, autonomous network topology structure information and the like) of the whole AS 2 network.

(1-6) the SDN controller in the AS 2 acquires the information of the autonomous network in the jurisdiction, and selects a common shortest routing algorithm to select a route for the packet _ in data packet by using a routing management module and a routing calculation module.

(1-7) packet _ in is forwarded to a multimedia server h2 through an optimal path, h2 responds to a request to generate a response data packet, the response data packet is communicated with an SDN controller through a previously established security channel, and the SDN controller receives the response data packet and calls a QOS module.

And (1-8) the SDN controller executes flow preprocessing to guarantee the QoS route of the subsequently generated multimedia flow.

(1-9) the SDN controller feeds back the flow preprocessing result to the h2 server, and the h2 refuses to provide the multimedia service or generates a packet _ service data packet according to the result. If the h2 server refuses to provide the multimedia service, after a period of time (Timeout), the h1 resends a new packet _ in, and repeats the above processes (1-2) to (1-8), otherwise, the step (1-10) is entered.

(1-10) h2 sends the packet _ service to the SDN controller, the controller judges the flow type of the packet _ service, sends the packet _ service back to the server h2 after the header of the packet _ service data packet is marked with the flow, and forwards the packet _ service in the autonomous network AS 2 according to the following rules:

if the packet _ service is a data stream, forwarding according to a general shortest path algorithm.

If the packet _ service is a media stream, forwarding by using a QOS routing algorithm.

And (1-11) installing corresponding flow table entries by the SDN controller through a secure channel along the route traveled by the packet _ service in the forwarding process (reversely installing the entries from S4 to S1) so as to ensure the consistency of the whole network resources.

(1-12) each time a period of time passes, the SDN controller recalls the topology module to acquire the real-time updated autonomous network information.

And (1-13) the Packet _ service flow reaches an edge network device (SDN switch) of the AS 2 autonomous network and is transmitted to the AS 1 autonomous network.

2. QoS register server transmission flow based on SDN framework:

network service requirements: h1 sends out video request, h2 completes video registration service, and network provides QoS guaranteed network connection.

(2-1) h1 sends out a video request, generates a request packet, generates a packet _ in at S1, and sends the packet _ in to an SDN controller at S1, and the controller performs routing and adopts a data flow routing algorithm (such as a shortest path algorithm to avoid other video flows);

(2-2) the controller installs the table entry (to ensure the consistency of the network), and inversely installs the table entry from S4 to S2;

(2-3) the viewing request is sent to h 2;

(2-4) h2 registers QoS service, h2 generates complete QoS application, h2 generates QoS Flow to h3, h3 communicates with controller to complete admission decision. If the direct admission is available, routing is carried out; if the network flow can not be directly admitted, the network flow is adjusted and then the route is selected;

(2-5) h3 informing h2 and h1 of the video service QoS registration result;

(2-6) h2 sending the video stream to h1, and reversely installing the table entry when the video stream reaches h 1;

(2-7) normal communication between h1 and h2 under the supervision of the SDN controller.

While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.

Claims

1. An SDN controller for accessing service transmission comprises an SDN controller interface, a network topology management module, route management, a flow management module, a route calculation module, a call forbidding module, a transmission traffic control module, a device management module, a network measurement module, a QOS module and an SDN standard control module;

wherein the SDN controller interface comprises:

a controller-service interface for a service provider to set management rules;

a call barring module: when the requested QoS parameters have no feasible routing route, the module rejects/blocks the current request and informs the SDN standard control module to execute a corresponding instruction;

the QOS module acquires the definitions of different data flows of a user and distinguishes the QoS service requirements of the data flows through the definitions; the SDN standard control module is used for generating a flow table, establishing safety information and issuing the flow table;

2. The SDN controller dynamic differentiation control method according to claim 1, comprising the steps of:

step 1: QOS flow pre-processing in the network:

judging whether the output port is idle, if so, adopting an absolute priority scheduling mode; otherwise, selecting the path with the lightest load for the small flow in the high-priority queue to forward until the output port is idle;

constructing a network topology structure diagram based on the network resource information, wherein the expression mode of the network topology structure diagram is an acyclic directed graph;

and respectively counting the flow quantity in the forwarding queue of each hop switch port in each path of the K paths based on the network resource information, accumulating the flow quantity in the forwarding queues of all hops to obtain the total flow quantity of each path, and taking the path with the minimum total flow quantity in the K paths as an optimal path.