Port-Based Rate Limiting, Policing, and Shaping

The switch QoS implementation supports the following two features for bandwidth management and traffic control:

• ingress port–based rate limiting—a mechanism to limit the traffic rate accepted by the specified ingress port

• egress port-based shaping—the process by which the system delays and transmits packets to produce an even and predictable flow rate

  Each port has eight unicast and multicast queues, Class of Service (CoS) 0 to CoS 7. Traffic shaping exists on the egress CoS 6 and CoS 7, but you cannot change the configuration. CoS 6 and CoS 7 are strict priority queues, with traffic shaping for CoS 6 at 50 percent and CoS 7 to five percent of line rate.

Each feature is important to deliver DiffServ within a QoS network domain.

Some hardware platforms support an ingress flow-based policer for ACLs. For information about Ingress Flow-based Policer, see Ingress Bandwidth Rate Limiter.

Token Buckets

Tokens are a key concept in traffic control. A port-based rate limiter, policer, shaper, or an ingress flow-based policer calculates the number of packets that passed, and at what data rate. Each packet corresponds to a token, and the port-based rate limiter, policer, shaper, or an ingress flow-based policer transmits or passes the packet if the token is available. For more information, see Token flow.

The token container is like a bucket. In this view, the bucket represents both the number of tokens that a port-rate limiter, policer, or shaper can use instantaneously (the depth of the bucket) and the rate at which the tokens replenish (how fast the bucket refills).

Each policer has two token buckets: one for the peak rate and the other for the service rate. The following figure shows the flow of tokens.

Token flow