Token bucket and bandwidth limits

 

Token bucket limits and bandwidth limits are together known as performance limits. These performance limits help guarantee the packet delivery in outbound bandwidth policies, both integrated and differentiated service.

 

Token bucket size

The token bucket size determines the amount of information that your system can process at any given time. If an application is sending your system information faster than the system can send the data out of the network, the buffer fills up. Any data packets exceeding this limit are treated as out-of-profile. Integrated service policies are the exception to this rule. You can select do not limit, which allows a ReSerVation Protocol (RSVP) connection request. For all other policies, you can determine how to handle out-of-profile traffic. The maximum token bucket size is 1 GB.

 

Token rate limit

The token rate limit specifies the long-term data rate or the number of bits per second allowed into a network. The quality of service (QoS) policy looks at the requested bandwidth and compares it with the rate and flow limits for this policy. If the request causes the system to exceed its limits, the system denies the request. The token rate limit is only used for admission control within the integrated service policies. This value can vary between 10 kbps to 1 Gbps. You can also set this value to do not limit. When you assign do not limit to the rate, you are making the available resources the limit.

Tip: To determine what limits to set, you might want to run the monitor. Create a policy with an aggregate token rate limit that is large enough to collect most data traffic on your network. Then start data collection on this policy. The scenario about monitoring current network statistics shows one way to collect the total rates your application and network currently use. Using these results, you can reduce the limits appropriately.

To view real-time monitor data instead of a particular data collection, just open the monitor. The monitor gives real-time statistics on all active policies.

 

Parent topic:

Integrated service

Related concepts
Differentiated service Scenario: Monitoring current network statistics