Intellinet 560559 User Manual page 58

Web Smart Managed PoE Switches User Manual
Strict Priority Queueing is specially designed to meet the demands of critical services
or applications. Critical services or applications such as voice are delay-sensitive and
thus require to be dequeued and sent first before packets in other queues are
dequeued on a congested network. For example, four egress queues 3, 2, 1 and 0
with descending priority are configured on a port. Under the SP algorithm, the port
strictly prioritizes packets from a higher-priority queue over those from a lower-
priority queue. Namely, only after packets in the highest-priority queue are emptied
can packets in a lower-priority queue be forwarded. Therefore, high-priority packets
are always processed before those of less priority. Medium-priority packets are
always processed before low-priority packets. The lowest-priority queue is serviced
only when highest-priority queues had no packets buffered. Disadvantages of SP:
The SP queueing method gives absolute priority to high-priority packets over low-
priority traffic, so it should be used with care. The moment a higher-priority packet
arrives in its queue, servicing of the lower-priority packets is interrupted in favor of
the higher-priority queue; packets can also be dropped if the amount of high-
priority traffic is too great to be emptied within a short time.
Weighted Round Robin
2.
The WRR queue-scheduling algorithm ensures every queue receives guaranteed
service time by taking turns according to a schedule determined by weighted values.
Assume a 100-M port has four egress queues with four weight values w3, w2, w1
and w0 and the corresponding values 25, 15, 5 and 5 that reflect the proportion of
resources assigned to those four queues. The queue with the lowest priority can be
assured of receiving at least 10 Mbps bandwidth, which avoids the disadvantage of
the SP algorithm where packets in low-priority queues may not be served for long
periods of time. Another advantage of the WRR algorithm is that, though the
queues are scheduled in turn, the service time for each queue is not fixed. That is,
when a queue is emptied, the next queue is scheduled for service. Therefore,
bandwidth resources are more fully utilized.
58