Layer 3 Switching Terminology
Several factors have created significant confusion surrounding the subject of Layer 3 switching. Some of this bewilderment arises from the recent merging of several technologies. In the past, switches and routers have been separate and distinct devices. The term switch was reserved for hardware-based platforms that generally functioned at Layer 2. For example, ATM switches perform hardware-based forwarding of fixed-length cells, whereas Ethernet switches use MAC addresses to make forwarding decisions. Conversely, the term router has been used to refer to a device that runs routing protocols to discover the Layer 3 topology and makes forwarding decisions based on hierarchical Layer 3 addresses.
Because of the complexity of these tasks, routers have traditionally been software-based devices. Routers have also performed a wide variety of “high touch” and value added features such as tunneling, data-link switching (DLSw), protocol translation, access lists, and Dynamic Host Configuration Protocol (DHCP) relay.
Layer 3 switching is a term that encompasses a wide variety of techniques that seek to merge the benefits of these previously separate technologies. The goal is to capture the speed of switching and the scalability of routing. In general, Layer 3 switching techniques can be grouped into two categories:
- Routing switches
- Switching routers
As a broad category, routing switches uses hardware to create shortcut paths through the middle of the network, bypassing the traditional software-based router. Some routing switch devices have been referred to as router accelerators. Routing switches do not run routing protocols such as Open Shortest Path First (OSPF) or Enhanced Interior Gateway Routing Protocol (EIGRP). Instead, they utilize various techniques to discover, create, or cache shortcut information. For example, Multiprotocol over ATM was discussed in Chapter 10, “Trunking with Multiprotocol over ATM.” This is a standards-based technique that allows ATM-attached devices to build a virtual circuit that avoids routers for sustained flows of information.
Although Cisco obviously supports MPOA, it has developed another shortcut technique that does not require an ATM backbone. This feature is called Multilayer Switching (MLS), although many people (and Cisco documents) still refer to it by an earlier name, NetFlow LAN Switching (NFLS). MLS is discussed in detail during this chapter.
Do not confuse MLS with other shortcut Layer 3 switching techniques that are not standards-compliant (many of these use the term cut-through switching). Many of these other techniques quickly switch the packets through the network without making the necessary modifications to the packet (such as decrementing the TTL field and rewriting the source and destination MAC addresses). MLS makes all of same modifications as a normal router and is therefore completely standards-compliant.
Unlike routing switches, switching routers do run routing protocols such as OSPF. These operations are typically run on a general-purpose CPU as with a traditional router platform. However, unlike traditional routers that utilize general-purpose CPUs for both control-plane and data-plane functions, Layer 3 switches use high-speed application specific integrated circuits (ASICs) in the data plane. By removing CPUs from the data-forwarding path, wire-speed performance can be obtained. This results in a much faster version of the traditional router. Switching routers such as the Catalyst 8500 are discussed in more detail later in this chapter.
Although the terms routing switch and switching router seem arbitrarily close, the terms are actually very descriptive of the sometimes subtle difference between these types of devices. For example, in the case of routing switch, switch is the noun and routing is the adjective (you didn’t know you were in for a grammar lesson in this chapter, did you?). In other words, it is primarily a switch (a Layer 2 device) that has been enhanced or taught some routing (Layer 3) capabilities. In the case of a switching router, it is primarily a router (Layer 3 device) that uses switching technology (high-speed ASICs) for speed and performance (as well as also supporting Layer 2 bridging functions).
Routing switches are Layer 2-oriented devices that have been enhanced to provide Layer 3 (and 4) functionality. On the other hand, switching routers are primarily Layer 3 devices that can also do Layer 2 processing (like any Cisco router).
Of the variety of other switching devices and terminology released by vendors, Layer 4 and Layer 7 switching have received considerable attention. In general, these approaches refer to the capability of a switch to act on Layer 4 (transport layer) information contained in packets. For example, Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) port numbers can be used to make decisions affecting issues such as security and Quality of Service (QoS). However, rather than being viewed as a third type of campus switching devices, these should be seen as a logical extension and enhancement to the two types of switches already discussed. In fact, both routing switches and switching routers can perform these upper-layer functions.