Analysis of Three TCP/IP transmission modes (unicast, broadcast, and Multicast)

Analysis of Three TCP/IP transmission modes (unicast, broadcast, and Multicast)

The Multicast Technology of TCP/IP transmission mode is a type of TCP/IP transmission mode. Before discussing Multicast technology, let's take a look at TCP/IP transmission methods. There are three TCP/IP transmission modes: unicast, broadcast, and multicast. Unicast transmission: a separate data channel is required between the sender and each receiver. If a host transmits data to a small number of recipients at the same time, there is generally no problem. However, it is difficult for a large number of hosts to obtain the same copy of data packets. This will lead to a heavy burden on the sender, long latency, and network congestion. To ensure a certain level of service quality, you need to increase hardware and bandwidth. Multicast Transmission: It improves data transmission efficiency. Reduces the possibility of congestion on the backbone network. The host in the multicast group can be in the same physical network or from different physical networks (if supported by the multicast router ). Broadcast Transmission: broadcasts packets in the IP subnet. All hosts in the subnet will receive these packets. Broadcast means that the network delivers a packet to the subnet host, whether or not these hosts are willing to receive the packet. However, the broadcast range is very small and only valid in the local subnet, because the vro blocks broadcast communication. Broadcast Transmission increases the overhead of non-recipients. Ii. Multicast Technology
2.1 The principle of multicast technology multicast is a method that allows one or more senders (Multicast sources) to send a single packet to multiple receivers (one time, at the same time) network technology. The Multicast Source sends data packets to a specific multicast group, and only the addresses belonging to the multicast group can receive data packets. Multicast can greatly save network bandwidth, because no matter how many destination addresses there are, only a single packet is transmitted on any link of the network. It improves data transmission efficiency. Reduces the possibility of congestion on the backbone network. The host in the multicast group can be in the same physical network or from different physical networks (if supported by the multicast router ). 2.2 If IP Multicast Transmission is implemented on the premise of multicast technology, both the source and receiver of multicast and the lower-layer networks between the two must support multicast. This includes the following: * The TCP/IP of the host supports sending and receiving IP multicast; * The network interface of the host supports multicast; * There is a set of group management protocols for joining, leaving, and querying, that is, IGMP (v1, v2). * There is a set of IP Address Allocation Policies, the layer-3 IP multicast address can be mapped to the layer-2 MAC address. * application software supporting IP multicast is supported; * All routers, hubs, switches, TCP/IP stacks, and firewalls between multicast sources and receivers must support multicast. 2.3. multicast addresses are in multicast communication, we need two types of addresses: an IP multicast address and an Ethernet multicast address. The IP multicast address identifies a multicast group. Because all IP data packets are encapsulated in Ethernet frames, a multicast Ethernet address is also required. To make multicast work properly, the host should be able to receive unicast and multicast data at the same time, which means that the host needs multiple IP addresses and Ethernet addresses. The IP address scheme specifies an address range for multicast. In IPv4, It is a Class D address in the range of 224.0.0.0 to 239.255.255.255, class D addresses are divided into local link multicast addresses, reserved multicast addresses, and management permission multicast addresses. Local link address: 224.0.0.0 ~ 224.0.0.255 is used for LAN. The Router does not forward IP packets in this range. The reserved multicast address is 224.0.1.0 ~ 238.255.255.255, used for global or network protocols; management permission address: 239.0.0.0 ~ 239.00000000255 is used inside the organization to restrict the multicast range. The last 28 bits of class D addresses are not structured, that is, there is no network ID or host ID. A host that responds to an IP address multicast address constitutes a host group, which can span multiple networks. The number of members in the host group is dynamic. The host can join or leave a host group through the IGMP protocol. For details about how to apply an IP multicast address to an ethernet address, see. Because the IP multicast address is 5-bit tall, the ethernet address is not unique. A total of 32 IP multicast addresses are mapped to an ethernet address. 2.4 multicast protocol: multicast protocol mainly includes Group Management Protocol (IGMP) and Multicast Routing Protocol (dense mode protocol (such as DVMRP, PIM-DM), Sparse Mode protocol (such as PIM-SM, CBT and MOSPF) * The IGMP host of the group management protocol uses IGMP to notify the subnet multicast router and wants to join the multicast group; the router uses IGMP to query whether a host belongs to a multicast group in the local subnet. * When a host is added to a multicast group, it notifies the multicast router of its IP subnet through the "membership Report" message, at the same time, prepare your own IP module to start receiving data from the multicast group. If this host is the first host to be added to the multicast group in its IP subnet, the multicast router is added to the multicast distribution tree through routing information exchange. * Exit the multicast group in IGMP v1. When the host leaves a multicast group, it automatically exits. When a multicast router is scheduled (for example, 120 seconds), the "Membership Query" message is used to query the Group addresses of all hosts in the IP subnet (224.0.0.1, if a multicast group does not have any Members in the IP subnet, after confirming this event, the multicast router will no longer forward data of the multicast group in the subnet. At the same time, route information is exchanged to delete the corresponding multicast Router from a specific multicast group distribution tree. This method of quietly leaving without notice makes the multicast router know that the event of no member in the IP subnet has been delayed for a period of time, so in IGMP v2.0, when a host leaves a multicast group, it must notify the subnet multicast router. The multicast router immediately asks all Multicast Groups in the IP subnet, this reduces the latency of the system to stop multicast. * To forward multicast packets in a real network, the multicast routing protocol must run on each interconnection device. Multicast Routing Protocols can be divided into three types: dense mode protocol (such as DVMRP, PIM-DM), Sparse Mode protocol (such as PIM-SM, CBT) and Link Status Protocol (MOSPF ), the following describes how each protocol works. * Distance Vector Multicast Routing Protocol: DVMRP (Distance Vector Multicast Routing Protocol: DVMRP) DVMRP is extended by the unicast Routing Protocol RIP. Both use the Distance Vector Algorithm to obtain the network topology information, the difference is that RIP forwards data forward based on the route table, while DVMRP is based on RPF. In order to enable new multicast members to receive multicast data in a timely manner, dvflat uses the method of regularly sending data packets to all the LAN. However, this method causes a large number of route control data packets to spread, this part of the overhead limits the expansion of the network. On the other hand, DVMRP uses the number of hops as the Metering Scale, and its upper limit is 32 hops, which is also a limitation on the network size. Hierarchical DVMRP is proposed, that is, the multicast network is divided into regions. multicast in the region can be implemented according to any protocol, cross-region multicast is performed by the VBR under the DVMRP protocol, which greatly reduces the routing overhead. * Author of the Open Multicast Shortest Path First: MOSPF (Multicast Open Shortest Path First: MOSPF) Protocol