Redundant Routed Protocols

Today's computers, and their network interface cards (NICs), can simultaneously support multiple networking protocols. In theory, each end system in a network could support two or more very different routed protocols, such as IP and IPX, simultaneously. The machine's network interface card doesn't look for these network layer addresses in data frames on its LAN; the NIC looks only at MAC addresses. The routed protocol datagrams are encapsulated in MAC-addressed frames. As the framing is unwrapped by the end system's data link layer protocol, the datagrams are revealed. This allows the data link layer protocol to examine the network layer address and forward the encapsulated data to the appropriate routed protocols for further processing.

The use of multiple network layer protocols is completely transparent to the end systems' NICs, as well as to the various LAN devices. Each of these protocols would, however, have its own address, address architecture, and datagram-forwarding mechanisms. This means that networks would have to support two sets of internetwork addresses---one for each routed protocol. Impacted devices would be routers and Layer 3 switches (if there are any in your network). Fortunately, routers are designed to support multiple routed protocols. Therefore, your network can already simultaneously use dissimilar routed protocols throughout the entire network.

Your options for supporting redundant routing protocols are either full or partial redundancy. Full Redundancy

A network that uses fully redundant routed protocols requires each networking device to simultaneously run both routed protocols. Consider, for example, the case of a networked environment that is in the midst of a transition from Apple-based computers and networking to Microsoft's Windows NT. The original networked environment used the AppleTalk protocol suite. The new network operating system, NT, will be used with IP. In such a scenario, a reasonable migration path might be to run both protocols on all devices simultaneously. Such a scenario could be temporary, until a migration can be completed. Alternatively, such a solution could be permanent, as would be the case with monitoring the IPv4-addressed management ports of older Ethernet hubs in an IPv6 network.

Figure 14-1 demonstrates fully redundant routed protocols in a small network.

Figure 14-1: Fully redundant routed protocols.

Figure 14-1: Fully redundant routed protocols.

In this scenario, any device could communicate with any other end system using either AppleTalk or IP addresses. Ostensibly, each end-system's applications would be configured to use one or the other protocol. As the transition progresses, support for AppleTalk could be phased out.

0 0

Post a comment