Routing Considerations

When implementing VLANs, you'll need some type of route processor (RP). An RP is a device that can switch information either between logical subnets (VLANs) or physical subnets (as in the traditional router). If the RP is performing a traditional routing role, it could be switching packets between different LAN media types, such as fiber distributed data interface (FDDI), Ethernet, and token ring. For WAN connections, it provides access to ISDN, frame relay, ATM, and dedicated circuit networks.

The RP is the main system processor in a Layer 3 device. It contains the main CPU, the operating system software, and most of the system memory components. Its primary function is to maintain and execute the management functions that control the Layer 3 device, including any routing protocols. The RP can be either an internal or external device. An example of an external RP is a Cisco 3600 or 7600 Series router. An example of an internal RP is the Multilayer Switch Feature Card (MSFC) that's installed on the Supervisor Engine card in a Catalyst 6500 Series switch. For the purposes of this book, all these Layer 3 functions are referred to as RPs.

Before you sit down and start configuring your RPs, you'll first have to plan out your VLANs and configure them on your switches. During this VLAN planning process, keep the following items in mind:

• Your Layer 3 addressing scheme

• How many VLANs you have

• What types of traffic are moving between VLANs

• How much traffic is moving between VLANs

• What kind of redundancy is required

• Your choice of routing protocols

• Layer 3 convergence issues

• Load balancing Layer 3 traffic

The preceding list is important for the Switching exam.

It's important to point out one major difference between a Layer 2 and a Layer 3 device. If a Layer 2 device, such as a switch, doesn't know how to reach a destination, it will flood the frame. If a Layer 3 device, such as a router, doesn't know how to reach a destination, it drops the frame.

Client End Station Issues

When you're implementing a Layer 3 routing protocol, clients must be able to find a router that can move their information to the destination subnet or VLAN. The most common method of accomplishing this is by assigning a default gateway address to every client. The default gateway address represents the Layer 3 address of the RP, in the same subnet, that the clients will use. For end stations, clients normally acquire this dynamically, such as using DHCP in an IP-based network. For more critical services, this is typically physically assigned to the end station. One problem that arises with this approach is redundancy. If the end stations know about only one RP, and that RP fails, how can the network move the traffic to its final destination? Cisco has solved this problem with a proprietary protocol called the Hot Standby Routing Protocol (HSRP). This protocol is discussed in Chapter 7, "Availability and Redundancy."

Route Processor Issues

Because an RP needs to route between different subnets, the question arises as to how it will accomplish this. Traditionally, each subnet requires a physically separate interface on the RP. This is okay if you have only a few subnets. But as your network grows, you eventually run out of ports on your RP. Another downside of this approach is that interfaces on an RP are expensive. Depending on your traffic patterns, some VLANs might not have very much inter-VLAN traffic, thus underutilizing an expensive interface.

A better approach to solving this problem is to use an interface on the RP that supports trunking that is, one interface that supports multiple VLANs, such as ISL or 802.1Q. If you remember from Chapter 3, "VLANs, Trunks, and VTP," ISL and 802.1Q use an encapsulation/tagging mechanism that identifies which VLAN a frame originated from. This is needed by the RP so that it can correctly distinguish which VLAN the frame came from and forward the frame to the correct destination.

Chapter 2, "Designing Switched Networks," discussed the roles of the three different layers of the campus model: access, distribution, and core. That chapter mentioned that VLAN boundaries need to be terminated at the distribution layer, thereby preventing broadcasts created in the access layers from propagating into the core layer and wasting valuable bandwidth. Therefore, your RPs should provide this wall at the distribution layer. To provide for the same traffic behavior and predictability, your RPs should have the same functionality and contain the same features. This also helps when you're performing your configuration and troubleshooting tasks by easing your administration of these devices. Your choice of RPs will be either external or internal.

An external RP would be a 2600, 3600, 7100, or some other type of router. An internal RP has both Layer 2 and Layer 3 components built into it, such as the Catalyst 3550 switch or the 6500 switch with the MSFC installed.