Analysis of multicast protocol and Multicast Route

Source: Internet
Author: User

1. What is multicast 1. What is multicast? Multicast is a data packet transmission method. When multiple hosts become the receiver of a data packet at the same time, multicast becomes the best choice for bandwidth and CPU load. 2. How does multicast work? Multicast uses the Class D address of as the destination address. One source host sends a packet whose destination address is the multicast address of the above range. In the network, if other hosts are interested in the packets in this group, you can apply to join the group and accept the group, others who are not members of this group cannot receive packets from this group.
3. What is the difference between multicast and unicast? In order for multiple hosts in the network to receive the same message at the same time, if the single-play method is adopted, the source host must continuously generate multiple identical messages for sending, for latency-sensitive data, the second data packet is generated when multiple identical data packets are generated on the source host, which is usually intolerable. It is also a huge burden for a host to continuously generate a message. If multicast is used, the source host can send only one message to reach each host that needs to be accepted. This is also dependent on the maintenance and selection of the relationship between the vro and the group members. 4. What is the difference between multicast and broadcast? As in the preceding example, when multiple hosts want to receive the same message, broadcast transmits the message to each host in the LAN regardless of whether the host is interested in the message. This will cause a waste of bandwidth and host resources. Multicast has a set of mechanisms for maintaining the relationship between Members and groups, so that you can clearly know whether hosts in a subnet are interested in such multicast packets, if no packet is sent, the upstream router is notified not to forward the packet to the downstream router.
Ii. Comparison between the elements of the multicast protocol and the data transmission modes of broadcast and unicast, we can find the two most important parts of Multicast: 1. in the multicast protocol, a set of protocols or mechanisms are required for the network device and the connected subnet to ensure that the network device knows the connected subnet, the number of hosts in a specific group. 2. What is the purpose of the multicast routing protocol for (1) It is the interface that discovers the upstream interface and is closest to the source. Because the multicast routing protocol only cares about the Shortest Path of the source. (2) through (S, G) pairs to determine the real downstream interface, when all routers know their upstream and downstream interfaces, then a multicast tree has been established. The root route is the router directly connected to the source host, and the branches are the routers that are directly connected to the subnet of the member through IGMP (3) Managing multicast tree unicast routes only need to know the next hop address, you can forward packets. Multicast sends a packet generated from the source to a group. On a specific vro, multiple backups of A package may be generated from multiple interfaces. If a loop exists, one or more packages will be returned to the input interface, and the package will be copied to other ports. This result may cause multicast storms. The packet is constantly copied between the vro and vswitch until the TTL is reduced to 0. Because this is a replication process, it has more serious harm than the unicast loop. Therefore, all multicast routers must know the source of multicast packets, ensure that multicast packets cannot be sent from the source interface. Therefore, he must know which upstream interfaces and downstream interfaces can distinguish the data packet flow. If the packet is not received by the source upstream interface, it is discarded. The Multicast Routing Protocol must concern the Shortest Path of the source, or the upstream interface of the source. Besides the upstream interfaces, data packets cannot be sent from interfaces other than the upstream interfaces during forwarding. Therefore, he also needs to care about (S, G) downstream interfaces. When the upstream and downstream interfaces of A (S, G) are determined, a multicast tree is formed.
Comparison between sparse and dense modes? What is sparse mode? It refers to a topology with a relatively small number of multicast hosts in a whole network, mainly in the WAN. What is intensive mode? In contrast to the above, what is the comparison between implicit addition and display addition in a switched LAN or campus network? A member can join or exit a group when the multicast session is alive, the connected vro must dynamically decide to add or remove branches of multicast trees based on the presence or exit of the members in the directly connected subnet. This is done by explicitly or implicitly adding two methods. Implicit addition is used in intensive mode. It uses broadcast/scissors mode to remove branches without multicast members on multicast trees. That is to say, it first adds all the routers on the Internet to the multicast tree, and then each router uses IGMP to check whether there are Members in the directly connected subnet. If no, A message is sent to cut out the excess branches. In Sparse Mode, each vro first queries whether there are Members in the subnet and then sends the information to the upstream vro.
Comparison between source-based and shared trees? The source-based tree is composed of a multicast tree for a single source. That is to say, if there are multiple source hosts in the network that can generate multicast packets, then there will be a number of multicast trees. In the multicast table, there will be a number of projects with the number of groups × number of members in each group. This topology is mainly applicable to intensive mode. The shared tree selects an RP or a centralized point in the network. All multicast messages need to be transmitted from this point, so it does not have (S, G) items, only (*, G) indicates that there are multiple sources. RP is a pre-defined router that is responsible for forwarding all multicast packets. All source hosts that want to send multicast packets must register on the RP before sending the multicast packets, and then use the directly connected vro to determine the Shortest Path of the RP, use the RP router to determine the shortest path to the destination. RP becomes the root node of the multicast tree. Compared with the source-based tree, the shared tree multicast table is more streamlined and suitable for use in Sparse Mode. But there are also some shortcomings. The selection of the Sharing tree on the RP will lead to the route from the source host to each group address is not the optimal path. If multiple bandwidth-consuming multicast links exist in the LAN at the same time, the RP becomes the bottleneck of the entire network. In addition, in the sharing tree, selecting RP to forward multicast packets increases the possibility of spof.
3. How to maintain group members: IGMP and CGMP (Cisco proprietary) Protocols to maintain the relationship between the members of the host and network devices. IGMP is the communication between the router and the internal subnet, that is, it is the maintenance mechanism of the layer-3 device for the group relationship of the directly connected subnet. It can be divided into two parts: the host part and the router part, each part can complete different work. However, it has a limitation that IGMP messages can only be transmitted within the local subnet, so that layer-3 devices cannot forward to other devices, and its TTL is always 1. IGMPv2 host function: the host running IGMPv2 generates the following three types of information: * The Member Report message is used to indicate that a host wants to join a multicast group, this message is sent when a host is added to the group for the first time. It can also be used to send a Membership Query message to a layer-3 device. Because the destination address of the Membership Query message is a group address, all the Members and hosts in the network will receive the packet except the vro. Once the other hosts receive the packet, they will suppress their own Membership Query Packets, this prevents the internal lan from flooding Membership Query Packets. It only needs to let the router know that there is another member in the network.
* Version 1 Membership Report messages are sent to IGMPv2 host backward compatibility and are used to detect and support IGMPv1 host and vro * Leave Group message host in the subnet, the destination address is (all routers), telling the router host to leave the group. IGMPv2 router function: is mainly a Query function. It has two types of Query Packets: General Query and Group-Specific Query General Query, which are sent to the LAN at intervals, the destination address is (all hosts in the network). Therefore, each host in the subnet receives this message and responds with the Member Report message, if the device does not receive any Member Report within a certain interval, it will think that there are no members in the subnet.
A Group-Specific Query packet. When a router receives a Leave Group packet, it sends a packet containing a Specific Group address to check whether a Group exists. If a sub-network has two multicast routers at the same time, they will initially think that they are the queryer of multicast members. When they send a General Query packet, compare the size of the source IP address received from the peer end to determine who is the queryer and who is not the queryer. If you do not receive the query message within a period of time, you will think that the queryer is down, and it will act as the queryer. CGMP (Cisco proprietary) considers that if a switch exists, it considers that there are Layer 2 devices in the layer 3 device and host, while IGMP is a Layer 3 protocol, if a layer-2 device receives such a packet, it will only forward the packet to all ports except the source port. This will affect the bandwidth and overall performance of the network. The solution is to allow the vswitch to forward multicast packets to the ports with multicast members. In a switched network, there are three methods to control multicast streams: (1) you can configure static multicast MAC address and port ing on the bridge table of the switch in the switched multicast tree. (2) GMRP (802.1 p) dynamically registers and cancels multicast addresses on the MAC layer on the vswitch. (3) By configuring IGMP listening on the switch port, the switch can perform IGMP message location check to know the multicast router and the location of the group members. However, detecting IGMP messages means that all local IPS must be checked. Especially if these are implemented in a software environment, the performance of the vswitch will be seriously reduced. The CGMP method is to use a vro to tell the vswitch the group MAC address and host MAC address of the multicast members so that the vswitch can know that there are Members on that port and forward the packets.
4. How to Use multicast routing protocols commonly used in Because mainstream router products only support Pim, other router protocols do not, therefore, we will only give a detailed explanation of PIM. DVMRP uses RIP to discover the shortest path to the source, and uses broadcast/scissors to construct a multicast tree. MOSPF discovers the shortest path to the source through the OSPF protocol, and is also used in intensive topology. CBT is a protocol-independent, sparse-mode-based, shared tree protocol. There are two differences between DVMRP and MOSPF: it does not need to add a Routing Protocol in multicast, and can find the shortest path of the source on any existing protocol; it is based on the shared Tree Protocol. Therefore, you must set a network core in the network to ensure the forwarding of multicast packets. It is more suitable for Sparse Mode. PIM-DM is a protocol-independent multicast protocol and a multicast protocol based on dense topology. The multicast tree is constructed using the broadcast/cut method.
Function: * detects neighbors by switching Hello packets. * When the unicast route changes, re-compute RPF interface * select a vro in a Multi-Channel Access Network * use scissors to overwrite in a Multi-Channel Access Network * use Assert messages in a multi-channel access network to elect a specified forward forwarder. generate the PIMv2 message in step 5: * The Hello message is periodically sent to identify the PIM neighbor. Similar to the Hello packet in OSPF, if the packet is not received within a certain period of time, the neighbor will be considered dead. * Join/Prune: This is a merged packet. The router information to be added to the multicast tree and the router information to be cut out are listed in this packet.
When the source is sending multicast messages, the PIM-DM uses the spread and cut method to establish multicast trees. Routers that do not have a member connection can send a Prune message to cut themselves out. When a router in the cut-off status receives the IGMP information from the directly connected subnet, it sends a join packet to the upstream router, the notification lists the subnets that need to be added to the multicast tree. * When a downstream router sends a Prune message to the upstream router, the upstream router does not immediately delete the advertised subnet from the multicast forwarding table, but waits for a while, if the subnet directly connected to the downstream router needs to be added to the multicast tree, a Graft message must be sent to notify the upstream router to add the multicast tree. * The upstream Graft-Ack router receives a Graft message and must return an Ack message, in this way, the downstream router receives the multicast packet * Assert. When two devices are connected to the same broadcast network at the same time and have the same upstream router interface, if the two routers are responsible for forwarding multiple texts at the same time, network resources will be wasted. Therefore, you can select one of the two routers as the only forward packet router, the Assert packet is used to communicate with each other in two vrouters, select a forward forwarder packet, and negotiate with the front forwarder through parameters such as priority value and IP address. PIM-SM uses the topology of the shared tree to establish multicast trees. Add the multicast tree by adding the display.
Find the aggregation point: * auto RP Protocol * Static settings PIM-SM and shared tree: PIM-SM because it is a Unidirectional Transport Protocol, multicast traffic can only be sent from the RP, instead, it cannot be sent to the RP, but the source sends the RP interface by encapsulating the multicast packets in the PIM Register packet, and then the RP unperforms the header, process the multicast packets. If there are many multicast packets, encapsulating Register packets is a burden on devices. Therefore, you can create a source-based tree between the RP and the source, send multicast packets from the source to the RP. PIM-SM and shortest path tree in a large network, the path of the shared tree is not necessarily the shortest path, the PIM-SM can achieve the Shortest Path transfer by establishing a source-based tree between the source and destination.

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.