IP Multicast

IP Multicast Basis

1-    Unicast: When transmitting with the unicast mode, the system creates a data transmission channel and sends a copy of the information for each subscriber who requested for the information. Thus, when a lot of users request for the same information at the same time, there is load of duplicate flows in the network. While unicast is a simple mechanism for one-to-one communication, for one-to-many communication it brings the network to its knees due to its huge bandwidth demands. When transmitting with the unicast mode one receiver receives one copy from the sender. The bandwidth of bottleneck is a big problem because the number of transmissions will increase as the number of users increases. So, the unicast mode is better applied to networks with few users instead of environments where messages are sent in large scales.

2-    Broadcast: In broadcasting, the system sends the information to all the users in a network whatever they need it or not, which undermines information security and impairs charging efficiency. Moreover, if only few receivers request the data, the network resources are used and the bandwidth is wasted. So, the broadcast mode is suitable for networks with a high density of users. If the number of users who request for the same information is not clear, neither the unicast nor the broadcast mode is a wise choice.

3-    Multicast: The information sender or the multicast source sends only one copy of the information to the network and creates tree-shaped routes by the multicast routing protocol to let the network intelligently replicate the packet only where it needs to, which helps conserve bandwidth and network resources both on the sending and the receiving end of a transmission. Multicast is based on the group. Receivers in one group use the same IP multicast address as the destination address. As shown in the diagram above, Receiver A and Receiver C joined the multicast group and then they can receive data. Compared with the unicast mode, multicast does not dramatically increase the burden of the network when the number of users hikes. Whatever how many receivers are there in the network, there is only one data stream with the same information. And unlike the broadcast scheme which requires the source to send an individual copy to each receiver, multicast conserves network resources by sending packets only to receivers who are interested in the information. In the multicast mode:The information sender is called the “multicast source”. Receivers that get the same information from a multicast group and each of the receivers are a “multicast group member”. Routers that provide the multicasting service are “multicast routers”. Multicast routers do not only provide multicasting services, but also carry out management on group members. Moreover, a multicast router itself can be a member of one or more multicast groups. And those groups do not have any physical or geographical boundaries—the hosts can be located anywhere on the Internet.

Advantages of multicast lie in:

·       Enhanced efficiency: it reduces network traffic and relieves the burden for servers and the CPU.

·       Optimized performance: it reduces traffic redundancy, saves network bandwidth and lightens the network load.

·       Distributed applications: it makes multipoint applications possible.

·       Multicast technology provides a satisfactory solution for transmitting information from a single point to multiple points which makes an efficient one-to many transmissions in IP networks a reality.

·       Multicast technology brings us versatile value-added services from on-line live broadcasting, Web TV, distance education, tele-healthcare to on-line radio broadcasting, and real-time audio/video conferencing.

Disadvantages of multicast:

·       Best effort: Since packet loss is inevitable, the multicasting application program should be designed to overcome packet loss instead of depending on the multicast network for reliability. “Reliable multicasting” is still under research.

·       No congestion avoidance mechanism: For lack of mechanisms like the TCP window mechanism and the slow start mechanism, multicast could cause traffic congestion on the network. If possible, multicasting application program is expected to avoid congestion by certain checking attempts.

·       Packet repetition: Some mechanisms of certain multicasting protocols such as the Asser mechanism and the SPT switch mechanism may cause packet repetition occasionally and the multicasting application program should tolerate that.

·       Packet disorder: Multicasting protocols may lead to packet disorder when packets reach their destination and the multicasting application program should take methods such as the buffer pool mechanism to rectify it.

Multicast IP address:

IANA has assigned the Class D address space to be used for IP Multicast. This means that all IP multicast-group addresses fall in this range: 224.0.0.0 -239.255.255.255. The Class D address cannot be put into the source IP address field of an IP packet.

Multicast IP models:

ASM (Any-Source Multicast):

In the ASM model, any information sender can be a multicast source to send information to a multicast address. Hosts can receive all the information sent to the multicast group after they join the group. In this model, receivers have no idea about where the multicast source is before they get information from the source and they can join or withdraw from the group at any time.

SFM (Source-Filtered Multicast):

The SFM model is similar to the ASM model as in both the two modes, the members of a multicast group are totally equal from the perspective of the information sender. However, the SFM model is an extension of the ASM model as it can filter multicast sources. In the SFM model, the upper layer software checks the source address of multicast packets and decides whether to let the packets from a specific multicast source to pass. Thus, receivers can only get data from some of the sources. To receivers, multicast sources are filtered and only some of them are valid.

SSM (Source-Specified Multicast):

In daily life, users are interested in information sent by particular sources and are reluctant to get information from other sources. The SSM model provides a service that enables users to choose sources at the client. The substantial difference between the SSM model and the ASM model is that in the SSM model, receivers get to know the position of the multicast source beforehand. The SSM model uses multicast addresses in a different range from the ASM model and sets up special forwarding paths between receivers and specific resources.

Multicast MAC address:

The MAC address of the destination is the MAC address of the receiver when packets are transmitted in the unicast mode on an Ethernet. However, when packets are transmitted in the multicast mode, the destination turns to a group of hosts instead of one receiver. In this case, we use the MAC multicast address. The MAC multicast address is used to label receivers that belong to the same multicast group at the data-link layer. As defined by IANA, the first 24 bits of the MAC multicast address is 0x01005e and the lower 23 bits are the lower 23 bits of the multicast IP address. 

Mapping from IP address to MAC address

The mapping of an IP address to its MAC address is illustrated by the above figure. The first 24 bits of a MAC multicast address is 0x01005e.The first four bits of an IP multicast address are 1110 which indicate that the address is a multicast address and in the last 28 bits, 23 bits of them are mapped to the MAC address.

Multicast correlative protocol

            Includes multicast management protocol for host registration and multicast routing protocol for selecting route and forwarding:

Multicast protocols consist of multicast management protocols for host registration and multicast routing protocols for route selection. Internet Group Management Protocol (IGMP) works between hosts and multicast routers. This protocol provides a mechanism for establishing and maintaining the relations between hosts and multicast routers. Multicast routing protocols are applied to multicast routers. These protocols create and maintain routes for multicasting and make packets forwarding more reliable and efficient. For the ASM model, multicast routing can be classified into the inter-domain routing and the intra-domain routing. Intra-domain multicast routing protocols are used to search multicast sources and build multicast trees to send information to receivers within autonomous systems. Intra-domain routing protocols include DVRMP, MOSPF and PIM.DVRMP stands for Distance Vector Multicast Routing Protocol and it is a protocol of dense mode. This protocol defines the maximum hops to be 32. MOSPF is an extension of OSPF and it supports multicasting by defining new LSAs. Protocol Independent Multicast (PIM) is a typical intra-domain multicast routing protocol and it comes in two modes, namely, the Dense Mode (DM) and the SM (Sparse Mode). DM is suitable when receivers are densely dotted in a network; while SM is fit for networks where receivers scatter sparsely. PIM must work together with unicast routing protocols. Inter-domain multicast routing protocols are used to transfer information between ASs. Multicast Source Discovery Protocol (MSDP) is able to spread news about multicast sources across ASs. Multiprotocol Border Gateway Protocol (MPBGP) is an extension of MBGP and it can advertise multicast routes across ASs. For the SSM model, there is no concept of the intra-domain and the inter-domain. As the receivers have the knowledge of where the multicast sources are beforehand, the router can create paths for multicasting directly with the function of PIM-SM.

Multicast distribution tree

The two basic types of multicast distribution trees ar:

•Source Path Tree

Source path tree is a multicast distribution tree with its root at the source and branches forming a spanning tree through the network to the receivers. Because this tree uses the shortest path through the network, it is also referred to as a shortest path tree (SPT). The network is responsible for building a tree for each source that sends multicast datagram.

Source path tree is a multicast distribution tree with its root at the source and branches forming a spanning tree through the network to the receivers. Because this tree uses the shortest path through the network, it is also referred to as a shortest path tree (SPT). The network is responsible for building a tree for each source that sends multicast datagram. The figure below shows an example of two STPs in a network since there are two multicast sources, namely, Source 1 and Source 2. R1 and R2 are the two receivers that get information from the two sources.

The two STPs are:

S1—A---C (R1) -----E (R2)

S2---F----D---C (R1) ------E (R2)

•Shared Tree:

Unlike source trees that have their root at the source, shared trees use a router as the root. This shared root is called a Rendezvous Point (RP) and all the shortest paths between the RP and its receivers form a distribution tree. When using a shared tree, there is only one tree in the network. All the sources in the network use the tree to send their traffic to the root and then the traffic is forwarded down the shared tree to reach all receivers.

In the example below, two sources, S1 and S2, share the tree D (RP) ----C (RT) ----E (R2).

Multicast data forwarding

In multicast routing, the source sends traffic to an arbitrary group of receivers that are represented by a multicast group address. The multicast router must determine which direction is upstream (towards the source) and which direction (or directions) is downstream and then forwards packets down the downstream. The concept of forwarding multicast traffic away from the source, rather than to the receiver, is called Reverse Path Forwarding (RPF).

Reverse Path Forwarding

RPF makes use of the existing unicast routing table to determine the upstream and downstream neighbors. A router only forwards a multicast packet when it is received on the upstream interface (RPF interface). RPF not only enables routers to correctly forward multicast traffic down the distribution tree, but also helps to guarantee that the distribution tree is loop free which is very important for multicasting. RPF detection is the essence of the RPF mechanism. When a multicast packet arrives at a router, the router performs the RPF detection on the packet. The packet is forwarded only after it passes the check. Otherwise, it is dropped.

For traffic flowing down a source tree, the RPF detection procedure works as follows:

·       Step 1. The router looks up the RPF interface of the multicast source in a SPT environment or the RPF interface of the RP when a shared tree is used. The RPF interface is the egress where the router sends out the packet.

·       Step 2. If a packet is received from the RPF interface, then it passes RPF detection and is forwarded.

·       Step 3. If the RPF detection in Step 2 fails, the packet is dropped.

In the figure above, Router RTE performs RPF detection after it received a multicast packet from S0. The RPF detection discovers that the packet is received from a wrong interface, so Router RTE discards the packet.

Multicast forwarding RPF example:

The process of a RPF detection is actually a process of consulting the unicast routing table. Routers look up the unicast routing table to see if the egress to the source is the interface that receives packets after they get multicast packets. If the egress to the source is not the interface that receives packets, then RFP detection fails and packets is dropped. In the example above, the S0 interface of the router enables RPF detection after it receives a multicast packet. After consulting the unicast routing table, it finds the egress to the source 151.10.0.0/16 is S1 which is not the interface that received the packet. So, it decides to discard the packet since the packet is not received from the right interface.

In this example that has the same network with the previous one, the router performs RPF detection after it receives a multicast packet. After looking up the unicast routing table, the router finds the egress to the source is S1 which is also the ingress of the packet. So, the packets passed the detection and is forwarded along the distribution tree.

END!