The Advantages of Multicast
Any form of network communication involving the transmission of information to multiple recipients can benefit from the bandwidth efficiency of multicast technology. Examples of applications involving one-to-many or many-to-many communications include video and audio broadcasting, videoconferencing / collaboration, stock quotation and news feed dissemination, database replication, software download and website cache.
To understand the efficiency of multicast, consider providing a single-channel video server, as shown in Figure 1. For full dynamic full-screen viewing, the video stream requires a server of about 1.5 Mbps to the client bandwidth. In a unicast environment, the server must send a separate video stream to the network for each client (this consumes a link bandwidth of 1.5 x n Mbps, where n = the number of client viewers). With the 10 Mbps Ethernet interface on the server, only six to seven servers to the client stream can fully meet the network interface requirements. Even with a highly intelligent Gigabit Ethernet interface on a high-performance server, the actual limit will be 250 to 300 video streams of 1.5 Mbps. Therefore, the server interface capacity may be an important bottleneck, limiting the number of unicast video streams per video server. The unicast transmission of replication consumes a large amount of bandwidth in the network, which is another important limitation. If the path between the server and the client traverses the h3 router hop and the h2 switch hop, the “multi-unicast” video consumes a router bandwidth of 1.5 x n x h3 Mbps, plus a switch bandwidth of 1.5 x n x h2 Mbps. As shown in Figure 1,100 clients are separated from the server by two router hops and two switch hops, and a single multi-unicast channel will consume 300 Mbps of the router bandwidth and 300 Mbps of the switch bandwidth. Multi-unicast will consume 20 Mbps of the router and switch bandwidth even if the video stream bandwidth is reduced to 100 Kbps (acceptable quality is provided in the smaller window on the screen).
In a multicast environment, the video server needs to transmit a single video stream for each multicast group, regardless of the number of clients to view it. The video stream is then replicated in accordance with the networkundefineds multicast router and the switchundefineds requirements to allow any number of clients to subscribe to the multicast address and to receive the broadcast. In a router network, replication occurs only on the branch of the distribution tree, so substantially all of the replication occurs at the last switch. In a multicast scenario, only 1.5 Mbps of the server is used to network bandwidth, and the remainder can be used for additional channels of other uses or video content. In the network, the multicast transmission provides similar efficiency, consuming only 1/ n of the bandwidth of the multi-unicast solution.
Obviously, in the case of a large number of recipients of replication transmission, even in a simple network with a small number of router and switch jumps, there are great differences in server load and network load. Additional functions of multicasting are useful in specific applications such as financial services. Multicast transmission is delivered to almost all members of the recipient group at the same time. The variability of delivery time is limited by the difference of end-to-end network delay in the range of server-to-client path. In unicast scheme, the server sorts by transmitting multiple copies of data, so the delivery time is very variable, especially for large transmission or large distribution list. Another unique feature of multicast is that the server does not know the unicast network address of any particular receiver transmitted-all recipients share the same multicast network address, so they can join the multicast group while keeping it anonymous.
Primer on Multicast Technology
Multicast transmission technology can be used in data link layer (layer 2) and network layer (layer 3). For example, Ethernet, optical fiber distributed data interface (FDDI) and SMDS all support unicast, multicast and broadcast MAC layer addresses. As a result, a single computer on these networks can listen to unicast addresses, multiple multicast addresses, and broadcast addresses at the same time. Token rings also support the concept of multicast, but use different techniques to handle receiver groups.
If the scope of the multicast application is limited to a single physical or logical lan, it is sufficient to multicast over the data link layer. However, most multi-point applications will be of concern only if you can extend to a distributed campus or even a wide-area environment that includes many different network technologies such as Ethernet, FDDI, Token Ring, Frame Relay, and ATM. For these extended environments, multicast must also be implemented at Layer 3. The multicast transmission of Layer 3 involves the following special mechanisms:Addressing-must have a layer 3 address to communicate with a set of receivers rather than a single receiver. In addition, there must be a way to map the address to the layer 2 multicast address of the underlying physical network. For IP networks, Class D addresses have been set aside for multicast addressing. Class D address is composed of 1110, which is the higher order bit in the first eight-bit byte, followed by the unstructured 28-bit group address. In order to map the IP multicast address to the Ethernet address, the low 23 bits of the Class D address are mapped to the Ethernet address block that has been reserved for the multicast. Using this mapping scheme, each Ethernet multicast address corresponds to 32 IP multicast addresses. This means that hosts that receive multicasting may need to filter out unwanted multicasting packets that are forwarded to other groups with the same MAC layer multicast address. The Ethernet multicast address has “01” in the first byte of the destination address to allow the network interface to easily distinguish between multicast and unicast packets.
Dynamic registration-there must be a mechanism to notify the network computer that it is a member of a particular group. Without this information, the network will be forced to flood instead of multicasting the transmission of each group. For IP networks, Internet group multicast protocol (IGMP) is a IP Datagram protocol between routers and hosts, which allows dynamic maintenance of group membership lists. The host sends a IGMP report or joins to the router to join the group. Routers regularly send queries to see which hosts are still part of the group. If the host wants to continue his group membership, it will use the report to respond to the query. If the host does not send a report, the router prunes the list of groups to minimize unnecessary traffic. With IGMP V 2, the host can send a “leave” message to inform the router that it is no longer involved in the multicast group. This allows the router to trim the list of groups before scheduling the next query, thereby minimizing the amount of time wasted transmission is forwarded to the network.
Multicast forwarding-most IP multicast applications are based on UDP, which uses “best-effort delivery” and lacks TCP congestion avoidance window mechanisms. As a result, multicast packets may be discarded more frequently than unicast TCP packets. Because it is impractical for real-time application request retransmission, audio and video broadcasting may be degraded due to packet loss. Before deploying quality of service (QoS), the best way to minimize the loss of packets in frame-based networks is to provide sufficient bandwidth, especially at the edge of the network. When the reservation protocol (RSVP), real-time transport protocol (RTP) and 802.1p or other layer 2 priority mechanisms can provide end-to-end QoS on layer 2, the reliability of multicast transmission / layer 3 network can be improved.
Multicast routing-the network must be able to build a packet distribution tree that specifies the unique forwarding path between the source subnet and each subnet that contains members of the multicast group. The main goal of the distribution tree construction is to ensure that a copy of each packet is forwarded at most on each branch of the tree. This is achieved by building a spanning tree with the designated multicast router as the root of the sending host, thus providing the connection to the specified multicast router for each receiving host. For IP multicast, IETF provides several multicast routing protocols for consideration. These include: (MOSPF), protocol independent (MOSPF), protocol for distance vector multicast routing protocol (DVMRP), OSPF and core-based tree (CBT). The multicast routing protocol constructs the distribution tree by checking the routing table of the unicast reachability protocol. Some protocols use unicast forwarding tables, including PIM and CBT. Or, other protocols use their own private unicast reachability routing tables. DVMRP uses its own distance vector routing protocol to determine how to build a source-based distribution tree. Similarly, MOSPF uses its own link-state database to build a source-based distribution tree.
There are two types of multicast routing protocols: dense mode (DM) and sparse mode (SM). The DM protocol assumes that almost all routers in the network need to allocate multicast traffic to each multicast group (for example, almost all hosts on the network belong to each multicast group). Therefore, the DM protocol builds the distribution tree by initially flooding the entire network and pruning a small number of paths without a receiver. The SM protocol assumes that relatively few routers in the network will participate in each multicast. Hosts belonging to this group are widely dispersed, as most multicasting on the Internet may be the case. Therefore, the SM protocol starts with an empty distribution tree and adds branches only as an explicit request to join the distribution. DM protocol, MOSPF,DVMRP and PIM-DM are most suitable for LAN environment with dense cluster receiver and bandwidth to tolerate flooding, while SM protocol (CBT and PIM-SM) is usually more suitable for WAN environment. PIM can also work in sparse dense mode by adjusting its behavior to match the characteristics of each receiver group.
If you are interested in the IP Multicast technology, and you can contact SPOTO: https://cciedump.spoto.net/ where there are the latest news about Cisco being updated.