Excess-Rate

Here you enter the bandwidth the Queue can use in addition to its Commit-Rate. The value is also commonly referred to as EIR (Excess Information Rate). To prevent higher-priority queues from taking the commit rate of lower-priority queues, the following concept was used:

The QoS operates in time slots, during which each queue can use its commit rate. At the end of the time slot, the unused Commit-Rate from all queues is carried over into the next time slot and used as a pool for the Excess-Rate. This pool then limits the bandwidth that can be used with the Excess-Rate. This fulfills two important aspects: Firstly, the Excess-Rate of a queue is not subtracted from another queue’s current Commit-Rate, but from the unused rate of the previous time slot. Second, the pool for the Excess-Rate is reset at the beginning of each time slot and is not added up, which means the unused Commit-Rate of a time slot can only be used in the following time slot. This prevents an accumulation, which could cause queues with a configured excess rate to starve the lower-priority queues.

Example: Two queues are configured, concatenated into a Queue-List, and assigned to a WAN interface. Queue A has a commit rate of 10 Mbps and an excess rate of 4 Mbps. Queue B has a commit rate of 5 Mbps and an excess rate of 0. If now in time slot 1 Queue A uses 9 Mbps and Queue B uses 4 Mbps, then 2 Mbps are unused rate and added to the pool of the excess rate for time slot 2. In this time slot, Queue A can then use its 10 Mbps commit rate and an additional 2 Mbps from the pool as part of its excess rate. Important is that only as much Excess-Rate can be used as the pool provides.