Learning Method:
Why
What
How
Why does this technology come into being. Without this technology, what is the current situation?
What is this technology? Its principle, operating process, its working process, its functions, and so on
How to Implement --- basic configuration, optimized configuration, advanced configuration, security features and summary.
Advantages of Multicast:
One-to-multiple, send data to multiple receivers at the same time.
Bandwidth saving
Saves hardware resources and so on.
Vro with IP multicast capability replicates multicast packets
The demand for bandwidth is constant, and unicast increases as demand increases.
Since multicast is so good, why do we need unicast?
Disadvantages of Multicast: UDP-based forwarding with best effort, no connection, and no concern about whether the receiver can receive the message. There is no serial number, and there may be no sequence during transmission, and the traffic control function is also lacking. No window mechanism. This causes repeated multicast traffic.
Why cannot multicast be based on TCP? ---- Multicast applications mainly involve real-time traffic, online videos, and stock trading. Highly time sensitive. Real Time.
If it is based on TCP, packet loss will be re-transmitted. If it is to watch a movie, the picture will be regressed because the re-transmission is completed. If it is UDP, It is skipped.
TCP transmission costs are high, and the header is large. The UDP cost is low, and the header is only 8 bytes. When transmitting real-time traffic, the lower the latency, the less the cost.
Multicast Application Type: 1-to-many, many-to-many, and many-to-1
One-to-many: One-to-multiple delivery; the other two are the same
IP multicast Compilation: Class D address 224 ~ 239. Like unicast, some reserved addresses are available and cannot be purchased. 224.0.0.0-224.0.0.255 is used locally and will not be routed to other network segments. It is mainly used for routing protocols ..
224.0.0.1 --- all hosts on the current subnet.
224.0.0.2 --- all the host routers on the subaccount.
224.0.0.5, 6, 9, 10---ospf rip OSPF
There are also private addresses: 239.0.0.0/8
Site-Local: 239.255.0.0/16
Organization-Local:
In addition to local management and retained IP addresses, they can all be called public addresses.
224.2.x.x
Unicast cannot be separated from multicast, but cannot be separated from unicast.
The well-known multicast address, no matter how long the TTL is and the TTL is, can only be transmitted in the local network.
Q: Why do we need to establish a physical direct connection between the neighboring nodes of the OSPF network in the past for the OSPF network. ----- Because their neighbor relationship is established through the hello mechanism, the hello packet is sent through multicast, And the unicast response is used to establish the neighbor relationship. The TTL value of the well-known multicast address is 1, so it can only be transmitted within the subscriber network. Therefore, you must establish a direct physical connection to establish a neighbor relationship.
ARP: ing between Layer 3 addresses and Layer 2 addresses
L2 multicast address: starting with 01-00-5e
Layer 3 rule, starting with 1110
The 28-bit Layer 3 maps to the 23-bit Layer 2. It means that an IP multicast MAC address represents 32 layer-3 multicast addresses.
Multicast session basics:
After the receiver is up, the available multicast session stream is obtained.
Methods for obtaining multicast session streams: email, RSS, and so on.
IP/TV: Source + Directory Server + Receiver
The receiver connects to the Directory Server through unicast to obtain the class table. Or listen to the program announcement to obtain the category table.
IGMP: maintain the relationship between multicast group members and groups.
V1, V2, and V3
Different from broadcasting: Before receiving traffic, you must add multicast groups and use the IGMP protocol.
Workflow: Query and response
V1: periodically sends a multicast query to 224.0.0.1 in 60 or 120 seconds. If a group contains a group member, the member sends a response to determine the number of multicast members. V1 has a large latency and exits implicitly. No IGMP Exit message. After the multicast router is queried three times, it will be known after 180 and no response is received. The group member has exited. If there are other members in the group, this will not affect much, because the multicast source still needs to send data to this group. If there are no other members in the group, it will take 180 seconds for the multicast router, to stop sending data to this group. Just like a teacher attending a class, students leave without prior notice.
V1. Similar to broadcast.
V2--RFC2236: the above two obvious shortcomings of V1 are improved.
1. query specific groups-when some groups change, if all members exit, the multicast router only sends a query to this group to save resources.
2. Exit Mechanism-send a leave message when a group member exits. Reduces latency.
When all the Members leave, the network segment will have no multicast traffic.
V3: optimized the source. Other mechanisms are similar to V2. The query response mechanism is the same.
When a host is added to a group, a message is immediately sent to 224.0.0.22. The information includes the multicast source that the host wants to receive multicast traffic and filters out the multicast source.
Show ip igmp interface --- view the IGMP version. IGMP V2 for Windows XP
Layer-2 Multicast: IGMP is a layer-3 protocol that is carried by IP addresses. If the value is 2 in the protocol field in the IP header, It is IGMP traffic. When a host sends a JOIN message, how does the router determine which host the switch is connected ?? Because the switch is layer-2 and IGMP is layer-3, SW ignores IGMP messages.
Implemented through cGMP and IGMP listeners
CGMP: Cisco private, IGMP listener: Industrial Standard.
The same purpose is to solve the layer-2 multicast problem caused by IGMP.
1. cGMP: both SW and vro run simultaneously. The multicast router sends the cGMP message to SW 01001_cdd.dddd. Information Content: Type field --- add or exit the MAC address of the + IGMP client and the MAC address of the multicast group
Bind the MAC address of the multicast group to the vswitch port.
2. IGMP listening: Allows SW to identify IGMP traffic. IGMP listeners require good device performance and intercept IGMP traffic, which consumes a lot of switch performance. If it is based on software, the CPU usage is high, but it is not based on ASIC.