IP multicast technology

In recent years, with the rapid development of information technology, a large number of network applications have increased, so that the existing huge data transmission volume has multiplied, and the optimized bandwidth can meet the needs of increasing data transmission volume, IP multicast technology is an important means to optimize bandwidth. It is applicable to data transmission services from multi-point to multi-point or from one point to multiple points. The basic principle is that IP multicast is completed based on the IP protocol; IP multicast forces the network to copy information packets at the forks of the data distribution tree, instead of repeatedly sending the same data packet by the information source node.
I. IP multicast model features
Global coordination is not required for user addition and removal. To add a multicast group, you only need to set an IP multicast address. In order to receive data, you can add a volume in the Special IP multicast intersection, without having to know the situation of other users in the group. The routing hides the details of multicast implementation for the user.
If multicast originated from the same source and ended different users with the same data, you need to define a multicast address so that the network can decide how to send the source data stream to the multicast address, how to organize data stream transmission on its link to make the best use of bandwidth.
A vro establishes a distribution tree to connect multicast group members and forward IP groups addressing multicast groups to networks with multicast group members, it also solves the loop problem in the multicast route selection.
The sender uses a multicast address to send a group. The sender only needs to know the address but does not know any information about the receiver. A group can have any source.
Ii. Select multicast routes
Multicast routing is a network problem. The solution to this problem requires complicated and sometimes difficult-to-handle protocols.
1. Two basic problems of multicast routing selection
1) target address
That is, how to let the network know each destination address of each group. Each vro is connected to a host with a multicast member. One member in the group sends multicast data. Each host uses the Internet multicast management protocol to communicate with directly connected routers.
When the Source sends multicast group to its vro1 1, The vro1 forwards the group to another vro2 2 in the network, vro2 2 forwards the group to vro3 3 and the target host connected to the route. The next hop router runs similar to this one. The result is that data streams from multicast sources are displayed on each host. If 3rd and 4th vrouters do not have any connected hosts as multicast members, 2nd vrouters should not be forwarded to other vrouters, as a result, the addresses not added to the multicast group also obtain information, and the target address is not clear. The solution is to run the Internet multicast management protocol to listen to the network to know which nodes on the network need to receive multicast communication protocols. IGMP has two functions: Reporting multicast member messages to the neighboring multicast routers, and building multicast groups for hosts and routes.
2) multicast routing loop problems
In a network with four vrouters, it is assumed that the router forwards data streams to all vrouters or interfaces of the host that require multicast data streams. Because the vrouters are connected to each other in addition to the host, the result is a multicast route selection loop, which completely consumes the bandwidth of the link in the transmission of data streams.
2. Reverse path forwarding
The basic idea of reverse path Forwarding is to build the Shortest Path returned to the source from each destination, except that multicast groups are forwarded to all interfaces without interfaces ). If the access port of the multicast group is not the Shortest Path returned to the source end, the group is discarded. Distance Vector multicast routing protocol is a widely used routing protocol for IP multicast. It uses reverse route forwarding to select routes for multicast. It works like this: when the vro receives a multicast group, check whether the group is on the interface that returns the source path in the shortest. If not, the group is discarded. If yes, the group is forwarded to the interface or vro.
1) First, the Multicast Source sends a group to vro1 1, and then vro1 1 forwards it to vro2 2 and vro4 4;
2) vro2 2 determines through its unicast route selection table that vro1 1 receives the packet on the Interface closest to the source, and then forwards the packet to the directly connected receiver, the recipient adds the group to the group using the Internet group management protocol. Vro2 2 is also forwarded to vro3 3;
3) Like vro4 2, vro4 4 uses its unicast route selection to determine that the group of vro1 1 is received on the Interface closest to the source. It then forwards the group to the connected receiver, which is added to the group using the Internet group management protocol. Vro4 4 also forwards the group to vro3 3;
4) vro3 3 receives the multicast groups of vro2 2 and vro4 4 and finds the route to select which vro is closest to the source host. Vro3 3 selects a channel connected to vro4 4, even if the distance from the channel connected to vro2 2 to the source host is the same, vro3 3 then forwards the multicast group to vro4 4 and directly connected destination host 2;
5) vro4 4 receives the multicast group of vro3 3, determines that the Group was not received on the interface of the most recent multicast source, and discards the group.
When applying IP multicast, users only need to send one copy of their own information to the IP multicast addresses of all receivers. This obviously best utilizes the network bandwidth, IP Multicast can also greatly reduce the burden on the sending server. IP Multicast can effectively transmit one-to-multiple data streams through the distribution tree. In this distribution tree, only one transmission path is required between two routers, data streams are copied only when the path is split.

When a host wants to join the group, the host sends an Internet multicast management protocol to the router that understands the multicast source, the vro sends an graft packet from the vector Multicast Routing Protocol to start receiving the necessary multicast communication. The graft packet is sent to the upstream stream until they find the distribution tree. This distribution tree transmits a multicast group to this group member. For the sender, you only need to simply send a single IP group with the corresponding multicast address as the destination address. The rest of the work is done by the router, the scale can be expanded from one receiver to several receivers. If the vro finds that all directly connected multicast hosts are out of the group and no downstream router is dependent on group communication, the vro sends a pruning message to the upstream stream to disable communication, the sender does not need to care about group member changes.

Related Articles]

• IP multicast technology and its application

• Several multicast protocols

• Solution to loop problems caused by multicast Static Routing