You now know that multicast networks behave differently from unicast networks. It is important to keep this in mind when planning the deployment of an IP multicast network. You should take several factors into consideration, including bandwidth implications, use of multicast applications, application requirements, user requirements, the location of the recipients, required equipment, cost, and, most importantly, what multicast source(s) will be used.
All these factors require attention and planning for a successful deployment of IP multicast throughout the network. You must also think upside down when thinking about multicast routing. As discussed in the preceding chapter, distribution trees are built based on the position of the root (source) of the tree. Therefore, when planning the routing for the multicast network, you must know where your sources or RPs will be located.
By taking the time to plan and prepare for a multicast deployment, you will avoid headaches later. You must become familiar with the customer's requirements as well as the effects that multicast will have on the existing network.
There are many methods of implementing multicast on a network. Commonly, institutions will want to connect with the Multicast Backbone (MBONE) multicast sessions; therefore, they must implement multicast through a Distance Vector Multicast Routing Protocol (DVMRP) tunnel or with Multicast Border Gateway Protocol (MBGP). If the multicast source is within the network and meant to stay within the confines of the network, other design issues come into play. It is important that you understand what each multicast routing protocol brings to the table when it comes to operational functionality.
By better understanding the many protocols and possible implementations of multicast, you will be able to better plan and prepare for its deployment. With so many options, there is bound to be a solution for almost any requirement. Through understanding requirements, and through preparing and planning, you can successfully implement an IP multicast network.
Part of deploying multicast is the determination of how much of the network should be multicast enabled. This is an important decision because it directly affects many aspects of multicast implementation. To strategically place the RPs, you must know where all the multicast leaf routers will be. Knowing the approximate number of potential multicast subscribers can have an effect on which protocols are run in the network to allow efficient multicast forwarding and routing.
The decision to use end-to-end deployment can be based on the applications that will be used or the intent of multicast implementation. If you are enabling multicast for a corporate application, you would need to enable multicast on every interface on every router throughout the enterprise. However, if you need to provide access to only the MBONE for the engineering department, or some other department within the organization, perhaps the most efficient method would not include end-to-end configuration and deployment.
It is important to keep in mind that the state of technology is dynamic. Today, you might receive a request from a single department for multicast access. Before jumping on the project and planning for just that department, consider that in the near future, it is likely that other departments will also request access. Applications that require end-to-end multicast capability might be purchased or integrated into the enterprise. It is far better to plan an end-to-end deployment and initially activate only the routers and interfaces that are needed than to plan your implementation on a limited initial activation. It is easier to 'build it right the first time' than to try to come back and work around or rebuild a poor IP multicast deployment.