Specifications

White Paper

Figure 1. Congestion causes in switches

A switch may be the fastest and largest non-blocking switch in the world, but if you have either of the two scenarios

shown in the above figure (as in just about every network in the world), then that switch can experience congestion.

At times of congestion, packets will be dropped if the congestion management features are not up to par. When

packets are dropped, retransmissions occur. When retransmissions occur, the network load can increase. In already

congested networks this can add to existing performance issues and potentially further degrade overall network

performance.

With converging networks, congestion management is even more critical. Latency sensitive traffic like voice and

video can be severely impacted if delays are incurred. Simply adding more buffers to a switch will also not

necessarily alleviate congestion problems either. Latency sensitive traffic needs to be switched as fast as

possible. First we need to identify this important traffic through classification techniques. Then we need to

implement buffer management techniques to avoid the higher priority traffic from being dropped during congestion.

Finally we need to incorporate scheduling techniques to forward important packets from queues as quickly as

possible. As you will read in the document, the Catalyst 6500 implements all of these techniques making its QoS

subsystem one of the most comprehensive in the industry today.

All of the QoS techniques described in the previous section will be explored in more detail throughout this paper.

3. Hardware Support for QOS in the Catalyst 6500

QoS support in the Catalyst 6500 is performed by Application Specific Integrated Circuits (ASIC’s) in hardware. This

is one of the major differentiators between the QoS capabilities on this platform and QoS capabilities on other Cisco

Router platforms where those router platforms execute most QoS functionality in software. The specific hardware

supporting QoS includes the Multi-layer Switch Feature Card (MSFC), the Policy Feature Card (PFC) and various

port ASIC’s on the line cards. The focus of this document will be on the QoS capabilities of the PFC and ASIC’s on

the line cards.

3.1 Policy Feature Card (PFC)

There are five versions of the Policy Feature Card in use today. The PFC3A , PFC3B, and PFC3BXL are integrated

into the Supervisor 720-3A, Supervisor 720-3B and Supervisor 720-3BXL respectively. The PFC3B is the only option

for the Supervisor 32, while the PFC3C and PFC3CXL are integrated into the Supervisor 720-10G-3C and

Supervisor 720-10G-3CXL.

There is a set of base QoS functions common to all versions of the PFC. Starting with the third generation of PFC

(PFC3A) supports the following:

●

Traffic policing using a dual leaky bucket algorithm.

●

Egress aggregate policer on both routed ports and VLAN switched virtual interface (SVI) ports

●

Traffic classification of ingress traffic using Access Control Lists to inspect information in the Layer 2, 3 and 4

headers

●

Re-marking of DSCP priority bits