CN1996931B

CN1996931B - Network multicast method

Info

Publication number: CN1996931B
Application number: CN2005100224906A
Authority: CN
Inventors: 罗向征; 尹茂
Original assignee: Maipu Communication Technology Co Ltd
Current assignee: Maipu Communication Technology Co Ltd
Priority date: 2005-12-31
Filing date: 2005-12-31
Publication date: 2010-09-22
Anticipated expiration: 2025-12-31
Also published as: CN1996931A

Abstract

This invention relates to one network set broadcast method in communication technique field, which comprises the following steps: 1a, generating tree topological structure and each point gets the structure; 1b, according to the structure, root point sends data from down degree of point. The invention has the following advantages of only in need of sending to certain next point in set broadcast point sending business message.

Description

Network multicast method

Technical field

The present invention relates to network communications technology field, relate in particular to the multicast service technology of internet mid point multiple spot.

Background technology

Universal day by day along with the network interconnection technology, the IP multicasting technology is realized point-to-multipoint communication efficiently because of it, saves Internet resources to greatest extent, more and more is subjected to people's favor.At present typical multicast application has: IPTV, video conference, distribution of information (card volume, Weather information etc.), multimedia remote education etc.Because also can't support the multicast of Internet scope at present, thus stride Internet multicast domain (support among the Internet multicast routing protocol localized network we be referred to as multicast domain) between can only be continuous by the clean culture tunnel.When striding internet developing multicast business at present, source node need send to other nodes by the tunnel with Business Stream, can't embody multicast and reduce the advantage that message duplicates.The bandwidth of supposing a video flowing is B, if N multicast domain arranged, only consider the output of point-to-multipoint between the territory, the bandwidth that also needs (N-1) * B, think that at present the video flowing of run-of-the-mill needs the bandwidth of 700K, when interlink node more for a long time, higher to the source node bandwidth requirement, the access bandwidth of general enterprise is difficult to satisfy high-quality multicast service demand, as video conference.

Summary of the invention

Technical problem to be solved by this invention is, a kind of network multicast method is provided, wide area network multicast particularly, by specific strategy, overcome the shortcoming of above prior art, striding foundation topology between the multicast domain of internet, each multicast domain in the topology is shared the transmission task that sends point, reduces the requirement of multicast service to bandwidth.

The scheme that the present invention solves the aforementioned technical problem employing is that network multicast method is characterized in that, may further comprise the steps: 1a, generation tree topology, and each node obtains this tree topology; 1b, according to tree topology, root node is to the downstream site distributing data.

Each node can obtain topology by manual configuration, also can be transmitted by main controlled node to obtain.

Further, among the described step 1a, tree topology is for calculate generating, and computational methods comprise the steps: 2a, according to amount of bandwidth, to all node sequencings; 2b, be radix with the lowest-bandwidth, with maximum downstream node number of each node of ratio calculation of bandwidth/radix; 2c, according to the maximum downstream node number of node sequencing and each node, put into downstream node successively step by step.

Among the described step 1a, tree topology is the topological structure of appointment.Root node among the described step 1b is the root node of appointment.Perhaps, the root node among the described step 1b is first node at the same level.

Further, among the described step 1a, each node independently obtains topological structure.Perhaps, before the described step 1a, 1A in steps also: set up the tunnel between non-main controlled node and other nodes, and specify main controlled node on each node, main controlled node sends topological structure by the tunnel; Main controlled node can directly send (1 jumps) by the tunnel herein, also can transmit by other nodes.

Perhaps, each node is set up the tunnel in twos, and specifies main controlled node on each node, and main controlled node sends topological structure by the tunnel.

Further, when each node carries out the multicast message forwarding,, revise the forwarding outlet of multicast message according to the topology of main controlled node issue.Perhaps, when certain node state changed, main controlled node recomputates topology immediately and to each non-main controlled node issue, non-main controlled node is revised the forwarding outlet of multicast message according to the topology of main controlled node issue.Described correction may further comprise the steps: 3a, all BPMT interfaces of removing from the outlet tabulation; 3b, judge whether the RPF interface is the BPMT interface, if not, think it oneself is source in the BPMT topology, be root node with oneself, search corresponding topology, add the lower exit of BPMT topology appointment; If the RPF interface is the BPMT interface, be that root node is searched topology then with RPF neighbours, add the lower exit that the BPMT topology is safeguarded.

The invention has the beneficial effects as follows,, when the multicast source node sends service message, only need send to specific next-hop node, share bandwidth pressure, reduced of the requirement of each node developing multicast business bandwidth by other nodes by the present invention.

The present invention is further illustrated below in conjunction with the drawings and specific embodiments.

Description of drawings

Fig. 1 sets up the schematic diagram of mode for a kind of tunnel of the present invention.

Fig. 2-4 is the schematic diagram of embodiment 1.

Fig. 5-6 is the schematic diagram of embodiment 2.

Embodiment

RPF as herein described (reverse transmission path forwarding) is an input function of router, and this function is used for checking the packet that router interface receives.

In the present invention, the internet access device of all multicast domain nodes should be supported multicast routing protocol and the preferential Multicast Tunnel of bandwidth of the present invention (BandWidth PreferenceMulticast Tunnel) mechanism.This paper will support access device of the present invention to be called the preferential Multicast Tunnel node of bandwidth (BPMT node).In business domains, the user at first will be at each configured tunneling technique (as Fig. 1) between node in twos, and specify a unique node as Master node (or claiming main controlled node), on other each BPMT node, specify the Master node then by configuration, and the internet access bandwidth of this BPMT node.Each node periodically carries out nodal information announcement to Master, mainly comprises this node bandwidth information, and notification time interval, node time-to-live etc.

If just part links to each other in twos, for example, see at non-main controlled node and main controlled node just and set up the tunnel that BPMT also can play a role.The MASTER node must link to each other with all nodes.

Master obtains the information of all nodes by advertised information,, calculates all internodal forwarding topologys (BPMT topology) and is distributed to all nodes according to following steps according to the access bandwidth of each node:

A, node is sorted by node bandwidth;

B, be radix with the lowest-bandwidth, bandwidth/radix obtains the maximum downstream node number of each node;

C, for each node calculates forwarding topology, be root with the specified node, by the bandwidth order all the other nodes are put into downstream node successively step by step.The child node number of each node is the downstream node number that calculated in the 2nd step.Passing rule with this, all nodes are put into forwarding tree, is the forwarding tree of root for each node forms with this node;

After d, Master have calculated topology, topology is distributed to all nodes, as the foundation of each node message forwarding.

Leave announcement or node when aging when Master receives, need recomputate forwarding topology and distribution immediately.

Embodiment 1:

With 6 nodes is example, and its forwarding bandwidth is respectively 1M, 1M, 2M, 2M, 4M, 4M;

At first pressing bandwidth to node sequencing, is radix with the lowest-bandwidth, and bandwidth/radix obtains the maximum downstream node number of each node, obtains following ordered series of numbers:

112244

Calculate topology for each node then, when calculating topology for the 1M node, with it is root, this node can only have 1 downstream node, put into the 4M node of bandwidth maximum by the bandwidth order, determine the downstream node of 4M node then, the 4M node can have 4 downstream nodes, put into remaining 4M, 2M, 2M, 1M node successively, obtain topology as Fig. 2.

When determining topology for the 2M node, with the 2M node is root, put into 2 downstream nodes successively, be all 4M, put into downstream node for first 4M node then, the 4M node has 4 downstream nodes at most, put into 2M, 1M successively, the 1M node gets final product, though second 4M node has the ability to arrange 4 downstream nodes, node has been arranged to finish, so need not to arrange any downstream node.Obtain topology as Fig. 3.

Same processing mode can obtain the topology of 4M node, as Fig. 4.

Each BPMT node can obtain all BPMT nodal informations in the business domains from the topology announcement of Master, the interface that links to each other with node is referred to as the preferential tunnel multicast interface of bandwidth (BPMT interface).When the BPMT node carries out the multicast message forwarding, BPMT topology according to the Master issue, revise the forwarding outlet of multicast message, at first from the outlet tabulation, remove all BPMT interfaces, judge then whether the RPF interface is the BPMT interface, if not, think it oneself is source in the BPMT topology, with oneself is root node, searches corresponding topology, adds the lower exit of BPMT topology appointment.If the RPF interface is the BPMT interface, be that root node is searched topology then with RPF neighbours, add the lower exit that the BPMT topology is safeguarded, may revise the RPF interface of transmitting according to the BPMT topology simultaneously is the BPMT interface of appointment in the BPMT topology.Source node only need send to service message the next-hop node of BPMT topology appointment, but not the every other node in the business domains is diffused into the whole service territory by next-hop node.Reduced bandwidth requirement to source node.

For example: (embodiment 2)

Referring to Fig. 5.Each BPMT equipment is the Internet access device (this figure ignores the network behind this equipment) of place network, and interconnected by the tunnel between each BPMT equipment, interconnected tunnel interface is called the BPMT interface, and each equipment is BPMT neighbours each other.All devices all moves multicast routing protocol (as PIM-SM, PIM-DM etc.) among the figure, suppose that source (S) is at BPMT equipment 1 place network, network behind other BPMT equipment has recipient (G), according to the multicast routing protocol routing, 3 BPMT interfaces on the BPMT equipment 1 all will be multicast service (S, G) outlet, this is the situation before BPMT revises, and supposes with BPMT equipment 1 to be BPMT forwarding topology such as Fig. 6 of root node.

BPMT equipment 1 forms when transmitting, to carry out the correction of forwarding interface according to the BPMT topology, at first remove total interface, whether the RPF neighbours that judge oneself then are BPMT neighbours, in this example be not, oneself to be root node inquiry BPMT forwarding topology, i.e. Fig. 6, the BPMT interface that will link to each other with BPMT equipment 2 adds multicast forward table, and multicast message just only can be forwarded to BPMT equipment 2.

BPMT equipment 2 is when formation is transmitted, according to the multicast route selection, to have the outlet of non-BPMT interface, when revising according to the BPMT topology, the RPF interface is the BPMT interface, be that root node is searched corresponding BPMT topology then with RPF neighbours' (BPMT equipment 1), be similarly as above topological diagram, topological diagram will add the forwarding interface tabulation with the BPMT interface that BPMT equipment 3 links to each other with BPMT equipment 4 in view of the above, multicast message will be transmitted to BPMT equipment 3 and 4 when being forwarded to own place network.

BPMT equipment 3 and 4 is when formation is transmitted, inquire about same BPMT topological diagram (BPMT equipment 3 and 4 RPF neighbours are similarly BPMT equipment 1), the BPMT interface of the RPF interface that renewal is transmitted for linking to each other with BPMT equipment 2 does not have downstream neighbor in the topological diagram, the outlet tabulation need not to revise.

After these processing, the multicast service message just will be forwarded to equipment 2 from BPMT equipment 1, be transmitted to equipment 3,4 by equipment 2 again, rather than directly be transmitted to equipment 2,3,4 by BPMT equipment 1.

In order to simplify application, specify Master by manual configuration, the information of collecting all nodes is to calculate suitable forwarding topology and to distribute.With regard to application, Master should be deployed in the node at multicast service server (as video server, videoconference server etc.) place, should be the equipment of stable, a high access bandwidth.When node no longer participates in multicast service (only general can appear under the situation of network operation), can select to leave, will send to Master and leave announcement this moment, leaves announcement and can trigger Master and upgrade forwarding topology and distribution again timely.

In order to prevent that illegal node from adding, the influence topology calculates even business is carried out, can be in the adding of Master by configuration ACL Control Node.

Claims

1. network multicast method is characterized in that, may further comprise the steps:

A, generation tree topology, each node obtains this tree topology;

B, according to tree topology, root node is to the downstream site distributing data;

The described tree topology of step a generates for calculating, and computational methods comprise the steps:

A1, according to amount of bandwidth, to all node sequencings;

A2, be radix with the lowest-bandwidth, with maximum downstream node number of each node of ratio calculation of bandwidth/radix;

A3, be root node,, all non-root nodes put into downstream node successively step by step according to the maximum downstream node number of node sequencing and each node with the specified node.

2. network multicast method as claimed in claim 1 is characterized in that, the root node among the described step a3 is first node at the same level.

3. network multicast method as claimed in claim 1 is characterized in that, among the described step a, each node independently obtains topological structure.

4. network multicast method as claimed in claim 1 is characterized in that, before the described step a, and 1A in steps also:

Set up the tunnel between non-main controlled node and other nodes, and specify main controlled node on each node, main controlled node sends topological structure by the tunnel.

5. network multicast method as claimed in claim 1 is characterized in that, before the described step a, and 1A in steps also:

Each node is set up the tunnel in twos, and specifies main controlled node on each node, and main controlled node sends topological structure by the tunnel.

6. network multicast method as claimed in claim 1 is characterized in that, when all nodes carry out the multicast message forwarding, according to the topology of main controlled node issue, revises the forwarding outlet of multicast message.

7. network multicast method as claimed in claim 1 is characterized in that, main controlled node upgrades topology and issues the topology that non-main controlled node is issued according to main controlled node, the forwarding outlet of correction multicast message to each non-main controlled node according to the variation of each node state.

8. as claim 6 or 7 described network multicast methods, it is characterized in that described correction may further comprise the steps:

3a, from outlet tabulation, remove the preferential Multicast Tunnel Interface of all bandwidth;

3b, judge whether the RPF interface is the preferential Multicast Tunnel Interface of bandwidth, if not, think it oneself is source in the preferential Multicast Tunnel topology of bandwidth, be root node with oneself, search corresponding topology, add the lower exit of the preferential Multicast Tunnel topology of bandwidth appointment; If the RPF interface preferential Multicast Tunnel Interface that is bandwidth is that root node is searched topology with RPF neighbours then, add the lower exit that the preferential Multicast Tunnel topology of bandwidth is safeguarded.