Catalyst 4500 Series Switch Software Configuration Guide, IOS XE 3.6.xE and IOS 15.2(2)Ex

CEF Features

CEF is advanced Layer 3 IP switching technology that optimizes performance and scalability for large networks with dynamic traffic patterns or networks with intensive web-based applications and interactive sessions.

CEF provides the following features:

• Improves performance over the caching schemes of multilayer switches, which often flush the entire cache when information changes in the routing tables.
• Provides load balancing that distributes packets across multiple links based on Layer 3 routing information. If a network device discovers multiple paths to a destination, the routing table is updated with multiple entries for that destination. Traffic to that destination is then distributed among the various paths.

CEF stores information in several data structures rather than the route cache of multilayer switches. The data structures optimize lookup for efficient packet forwarding.

Forwarding Information Base

The Forwarding Information Base (FIB) is a table that contains a copy of the forwarding information in the IP routing table. When routing or topology changes occur in the network, the route processor updates the IP routing table and CEF updates the FIB. Because there is a one-to-one correlation between FIB entries and routing table entries, the FIB contains all known routes and eliminates the need for route cache maintenance that is associated with switching paths, such as fast switching and optimum switching. CEF uses the FIB to make IP destination-based switching decisions and maintain next-hop address information based on the information in the IP routing table.

On the Catalyst 4500 series switches, CEF loads the FIB in to the integrated switching engine hardware to increase the performance of forwarding. The integrated switching engine has a finite number of forwarding slots for storing routing information. If this limit is exceeded, CEF is automatically disabled and all packets are forwarded in software. In this situation, you should reduce the number of routes on the switch and then reenable hardware switching with the ip cef command.

Adjacency Tables

In addition to the FIB, CEF uses adjacency tables to prepend Layer 2 addressing information. Nodes in the network are termed adjacent if they are within a single hop from each other. The adjacency table maintains Layer 2 next-hop addresses for all FIB entries.

Adjacency Discovery

The adjacency table is populated as new adjacent nodes are discovered. Each time an adjacency entry is created (such as using the Address Resolution Protocol (ARP), a link-layer header for that adjacent node is stored in the adjacency table. Once a route is determined, the link-layer header points to a next hop and corresponding adjacency entry. The link-layer header is subsequently used for encapsulation during CEF switching of packets.

Adjacency Resolution

A route might have several paths to a destination prefix, such as when a router is configured for simultaneous load balancing and redundancy. For each resolved path, a pointer is added for the adjacency corresponding to the next-hop interface for that path. This method is used for load balancing across several paths.

Adjacency Types That Require Special Handling

In addition to adjacencies for next-hop interfaces (host-route adjacencies), other types of adjacencies are used to expedite switching when certain exception conditions exist. When the prefix is defined, prefixes requiring exception processing are cached with one of the special adjacencies listed in Table 35-1 .

Unresolved Adjacency

When a link-layer header is prepended to packets, FIB requires the prepend to point to an adjacency corresponding to the next hop. If an adjacency was created by FIB and was not discovered through a mechanism such as ARP, the Layer 2 addressing information is not known and the adjacency is considered incomplete. When the Layer 2 information is known, the packet is forwarded to the route processor, and the adjacency is determined through ARP.

Hardware and Software Switching

For the majority of packets, the integrated switching engine performs the packet forwarding function in hardware. These packets are hardware-switched at very high rates. Exception packets are forwarded by the CPU subsystem software. Statistic reports should show that the integrated switching engine is forwarding the vast majority of packets in hardware. Software forwarding is significantly slower than hardware forwarding, but packets forwarded by the CPU subsystem do not reduce hardware forwarding speed.

Figure 35-2 shows a logical view of the integrated switching engine and the CPU subsystem switching components.

Figure 35-2 Hardware and Software Switching Components

The integrated switching engine performs inter-VLAN routing in hardware. The CPU subsystem software supports Layer 3 interfaces to VLANs that use Subnetwork Access Protocol (SNAP) encapsulation. The CPU subsystem software also supports generic routing encapsulation (GRE) tunnel.

Hardware Switching

Hardware switching is the normal operation for Supervisor Engine III and Supervisor Engine IV.

Software Switching

Software switching occurs when traffic cannot be processed in hardware. The following types of exception packets are processed in software at a much slower rate:

• Packets that use IP header options

Note Packets that use TCP header options are switched in hardware because they do not affect the forwarding decision.

• Packets that have an expiring IP time-to-live (TTL) counter
• Packets that are forwarded to a tunnel interface
• Packets that arrive with non-supported encapsulation types
• Packets that are routed to an interface with non-supported encapsulation types
• Packets that exceed the MTU of an output interface and must be fragmented
• Packets that require an IGMP redirect for routing
• 802.3 Ethernet packets

Load Balancing

The Catalyst 4500 series switch supports load balancing for routing packets in the integrated switching engine hardware. Load balancing is always enabled. It works when multiple routes for the same network with different next-hop addresses are configured. These routes can be configured either statically or through a routing protocol such as OSPF or EIGRP.

The hardware makes a forwarding decision by using a hardware load sharing hash function to compute a value, based on the source and destination IP addresses and the source and destination TCP port numbers (if available). This load sharing hash value is then used to select which route to use to forward the packet. All hardware switching within a particular flow (such as a TCP connection) is routed to the same next hop, which reduces the chance that packet reordering occurs. Up to eight different routes for a particular network are supported.

Software Interfaces

Cisco IOS for the Catalyst 4500 series switch supports GRE and IP tunnel interfaces that are not part of the hardware forwarding engine. All packets that flow to or from these interfaces must be processed in software and have a significantly lower forwarding rate than that of hardware-switched interfaces. Also, Layer 2 features are not supported on these interfaces.

Enabling CEF

By default, CEF is enabled globally on the Catalyst 4500 series switch. No configuration is required.

To reenable CEF, perform this task:

Configuring Load Balancing for CEF

CEF load balancing is based on a combination of source and destination packet information; it allows you to optimize resources by distributing traffic over multiple paths for transferring data to a destination. You can configure load balancing on a per-destination basis. Load-balancing decisions are made on the outbound interface. You can configure per-destination load balancing for CEF on outbound interfaces.

The following topics are discussed:

• Configuring Per-Destination Load Balancing
• Configuring Load Sharing Hash Function
• Viewing CEF Information

Configuring Per-Destination Load Balancing

Per-destination load balancing is enabled by default when you enable CEF. To use per-destination load balancing, you do not perform any additional tasks once you enable CEF.

Per-destination load balancing allows the router to use multiple paths to achieve load sharing. Packets for a given source-destination host pair are guaranteed to take the same path, even if multiple paths are available. Traffic destined for different pairs tend to take different paths. Per-destination load balancing is enabled by default when you enable CEF; it is the load balancing method of choice in most situations.

Because per-destination load balancing depends on the statistical distribution of traffic, load sharing becomes more effective as the number of source-destination pairs increases.

Use per-destination load balancing to ensure that packets for a given host pair arrive in order. All packets for a certain host pair are routed over the same link or links.

Configuring Load Sharing Hash Function

When multiple unicast routes exist to a particular destination IP prefix, the hardware sends packets matching that prefix across all possible routes, which shares the load across all next hop routers. By default, the route used is chosen by computing a hash of the source and destination IP addresses and using the resulting value to select the route. This preserves packet ordering for packets within a flow by ensuring that all packets within a single IP source/destination flow are sent on the same route, but it provides a near-random distribution of flows to routes.

You can change the load-sharing hash function. So, in addition to the source and destination IP addresses, the source TCP/UDP port, the destination TCP/UDP port, or both can also be included in the hash.

To the configure load sharing hash function to use the source and/or destination ports, perform this task:

Note The include-ports option does not apply to software-switched traffic on the Catalyst 4500 series switches.

Viewing CEF Information

You can view the collected CEF information. To view CEF information, perform this task:

Displaying IP Statistics

IP unicast statistics are gathered on a per-interface basis. To display IP statistics, perform this task:

This example shows how to display IP unicast statistics for fastethernet 3/1:

To display CEF (software switched) and hardware IP unicast adjacency table information, perform this task:

This example shows how to display adjacency statistics:

Note Adjacency statistics are updated approximately every 10 seconds.