Switching Paths

The switch is commonly referred to in marketing terms as a Layer 2 device. If you keep thinking that way, this section will confuse you. By definition, switching paths are logical paths that Layer 3 packets follow when they are switched through a Layer 3 device such as a router or internal route processor. These switching types allow the device to push packets from the incoming interface to the interface where the packet must exit using switching paths or table lookups. By using switching paths, unnecessary table lookups can be avoided, and the processor can be freed to do other processing.

You’re probably wondering, “Sean, this is a switching book. Why am I learning about switching paths in Layer 3 devices?” Well, inside switches are Layer 3 devices such as the Route Switch Module (RSM), Multilayer Switching Module (MSM), Multilayer Switch Feature Card (MSFC), and NetFlow Feature Card (NFFC). Later in this book, I will cover trunk links, which are links that carry more than one VLAN. Doesn’t it seem logical that if you need to have a “router on a stick,” which is an external router used for interVLAN routing, it might help to know if the router you are using can handle the traffic for all of your VLANs? Better yet, you should learn the internal working paths and types of switching paths through the route processor. Let’s take a look at all the switching paths used on Layer 3 devices.

In this section, we will focus on the following switching path types and the functions of each:

∙Process switching

∙Fast switching

∙Autonomous switching

∙Silicon switching

∙Optimum switching

∙Distributed switching

∙NetFlow switching

Process Switching

Process switching uses the processor to determine the exit port for every packet. As a packet that needs to be forwarded arrives on an interface, it is copied to the router’s process buffer, where the router performs a lookup based on the Layer 3 destination address and calculates the Cyclic Redundancy Check (CRC). Subsequent packets bound for the same destination interfaces follow the same path as the first packet.

This type of switching can overload the processor. Making Layer 3 lookups the responsibility of the processor used to determine which interface the packet should exit takes away from more essential tasks the processor needs to handle. It is recommended that you use other types of switching whenever possible.

Fast Switching

Consider fast switching an enhancement to process switching. This switching type uses a fast switching cache found on the route processor board. The first received packet of a data flow or session is copied to the interface’s processor buffer. The packet is copied to the Cisco Extended Bus (CxBus) and then sent to the switch processor. If a silicon or autonomous switching cache does not contain an entry for the destination address, fast switching is used because no entries for the destination address are in any other more efficient caches.

Fast switching copies the header and then sends the packet to the route processor that contains the fast switching cache. If an entry exists in the cache, the packet is encapsulated for fast switching, sent back to the switch processor, and then buffered on the outgoing interface processor.

78

Autonomous Switching

With autonomous switching, when a packet arrives on an interface, it is forwarded to the interface processor. The interface processor checks the silicon−switching cache; if the destination address is not contained in that cache, the autonomous cache is checked. The packet is encapsulated for autonomous switching and sent back to the interface processor. The header is not sent to the route processor with this type of switching.

Note Autonomous switching is available only on AGS+ and Cisco 7000 series routers that have high−speed controller interface cards.

Silicon Switching

Silicon−switched packets use a silicon−switching cache on the Silicon Switching Engine (SSE) found on the Silicon Switch Processor (SSP). This is a dedicated switch processor used to offload the switching process from the route processor. Packets must use the router’s backplane to get to and from the SSP.

Optimum Switching

Optimum switching is similar to all the other switching methods in many ways. As the first packet for a flow arrives on an interface, it is compared to the optimum switching cache, appended, and sent to the destination exit interface. Other packets associated with the same session then follow the same path. Just as with process switching, all the processing is carried out on the interface processor.

Unlike process switching, optimum switching is faster than both fast switching or NetFlow switching when the route processor is not using policy networking such as access lists. Optimum switching is used on higher−end route processors as a replacement for fast switching.

Distributed Switching

Distributed switching is used on the VIP cards, which use a very efficient switching processor. Processing is done right on the VIP card’s processor, which maintains a copy of the router’s own route cache. This is another switching type in which the route processor is never copied with the packet header. All the processing is off−loaded to the VIP card’s processor. The router or internal route processor’s efficiency is dramatically increased with a VIP card added.

NetFlow Switching

NetFlow switching is usually thought of as utilizing the NetFlow Feature Card (NFFC) or NFFC II inside the Catalyst 5000 or 6000 family of switches. These switches use the NFFCs to let a router or internal route processor make a routing decision based on the first packet of a flow. The NFFCs then determine the forwarding interface decision made by the router or internal route processor and send all subsequent packets in the same data flow to that same interface. This method offloads work that the router used to do on to the switch’s NFFC card.

However, NetFlow switching is not just a switching type; it can be used as an administrative tool to gather statistics in an ATM−, LAN−, and VLAN−implemented network. This type of switching actually creates some added processing for the router or an internal route processor by collecting data for use with circuit accounting and application−utilization information. NetFlow switching packets are processed using either the fast or optimum switching methods, and all the information obtained by this switching type is stored in the NetFlow switching cache; this cache includes the destination address, source address, protocol, source port, destination port, and router’s active interfaces. This data can be sent to a network management station for analysis.

79