Chapter 7: IP Multicast

In Depth

Over time, our use of computers has moved from the local desktop to the local network. Most of this network traffic consists of using local network sources, such as printing and file transfers. As our personal computers become more powerful, the applications that we use now contain more audio and video components. We’ve now started to move this traffic onto the network.

Today, it is still the norm for network traffic to be between one sender and one recipient. That is slowly changing. As different network resources become available, new resources such as messaging, multimedia, distance learning, and Internet access are causing a large increase in data traffic. This type of traffic usually involves one server sending a data stream to multiple users; a good example of this would be video conferencing and software updates in the intranet.

This type of traffic can be very demanding in terms of data usage. For example, if 100 people want a 1.5MB file, the result is a demand for more than 150MB of data−link usage. Even on a T3, that’s a lot of simultaneous use. What’s more, this calculation doesn’t take into account the rest of the users’ applications and data requirements.

One way to provide users with this high−bandwidth information and at the same time minimize the traffic on the network is to utilize IP multicast. IP multicast enables data to be sent once and received by all the recipients that requested it.

The concept behind IP multicasting is that end recipients join a multicast group. The information that is requested is then delivered to all members of that group by the network infrastructure. The sender of the data doesn’t need to know anything about the recipients. In this manner, only one copy of a multicast message will pass over any link in the network, and copies of the message will be made only where the paths diverge. This is a much more effective method of delivering traffic destined for multiple locations, and it provides significant performance improvements for the network.

In this chapter, we will explain the concepts behind IP multicasting. We will cover the types of multicast traffic and introduce you to the way multimedia traffic types are routed on the network. Finally, we will look at the methods to configure IP routing on your Catalyst switches and how to manage the resulting multicast traffic. We will begin with a discussion of the different types of multicast traffic.

IP Multicasting Overview

IP multicasting is an extension of the standard IP protocol and is described in RFC 1112, “Host Extensions for IP Multicasting.” IP multicasting is the transmission of an IP datagram to a group identified by a single IP destination address. A multicast datagram is delivered to all members of its destination host group using User Datagram Protocol (UDP). Membership in these groups is unrestricted—hosts can be members of multiple groups, and they may join or leave at any time.

IP multicast datagrams are handled by multicast routers. A host transmits an IP multicast datagram as a local network multicast that reaches a multicast router. The router examines the packet and begins to provide the host with the requested multicast traffic. If the router is not receiving the requested multicast traffic, it will pass the request to other multicast routers.

IP traffic can travel the network in one of three ways:

∙ Broadcast

137

∙Unicast

∙Multicast

Broadcast

In its simplest form, broadcast traffic consists of packets that reach every point of the network. In a typical network, broadcasts are stopped at the router. You can set the router to forward broadcasts, but doing so is not very efficient—it creates a lot of traffic on the network and slows the end users’ machines. Every host on the network must process the packet to see if it is destined for that host. Data broadcasts are typically small frames used in the local network—so, the performance effect is negligible, unless there is a broadcast storm.

Note In a broadcast storm, an incorrect packet is broadcast on the network. This causes most hosts to respond with incorrect answers, which in turn causes even more hosts to respond again. This process continues until the network can no longer carry any other traffic. A broadcast storm can also occur when there is more than one path through the network, allowing broadcasts to circle the network until there are so many that the network comes to a stop.

Multimedia broadcasts, in contrast, can be huge packets. Processing these types of broadcasts can quickly use up all the available bandwidth on the network and bring the end station to a crawl—particularly if you are in a shared 10BaseT environment.

Figure 7.1 illustrates broadcast traffic in the network.

Figure 7.1: : Broadcast traffic flow.

Unicast

In unicast, a single packet is sent from the source to the destination. It is a one−to−one relationship: For every packet that reaches the destination, one packet was sent by the source. This process is fine if the source is having different conversations with only a few hosts. Now, imagine that same source talking to hundreds of hosts on the same conversation—each identical packet must be generated by the source and must travel on the network.

Audio and video transmissions are so large that a high−bandwidth link is consumed very quickly. A 100Mbps link can support about 60 to 70 full−screen, full−motion video streams if each stream uses approximately 1.5Mbps of server−to−client bandwidth. You will need gigabit−per−second (Gbps) links between the server and the network in order to provide one audio/video broadcast to a couple hundred hosts. Unicast multimedia applications do not scale very well.

Figure 7.2 illustrates unicast traffic flow.

138

Figure 7.2: Unicast traffic flow.

Multicast

Multicast is a combination of broadcast and unicast. It sends one copy of the packet to many hosts that requested it, thereby using less bandwidth. It also saves bandwidth by not sending the packet to the portion of the network whose hosts didn’t request the transmission. Multicast accomplishes this task by transmitting to an identified group, called a multicast group, rather than to an individual host. Each interface/host can be a member of multiple multicast groups. The membership is dynamic; a host can leave and join any time it wants. The traffic is also not limited by any boundary; it can reach the farthest point of the Internet.

Figure 7.3 illustrates multicast traffic flow.

Figure 7.3: Multicast traffic flow.

The characteristics of multicast enable it to take three different forms:

∙One−to−many—One−to−many is the most common form of multicast traffic. Examples include database updates, live concerts, news, music/audio broadcasts, announcements, lectures, and many more.

∙Many−to−one—Many−to−one multicasts are less common; they include data collection, auctions, and polling.

∙Many−to−many—Many−to−many multicasts are rare, but they are gaining popularity as programmers begin to utilize multicast in some imaginative ways. Chat groups, multimedia

conferencing, concurrent processing, interactive music sessions, and collaboration are examples of many−to−many multicasts. But don’t forget the rising star (and my favorite): interactive multiplayer games.

Want to Join the Military?

The U.S. military has one of the largest interactive multicast−based war−game simulations I’ve heard of. The battlefield is divided into map grids, and each grid square is a multicast group. Individuals communicate with

139