QTECH
Software Configuration Manual
14-180
14.1.5
Flow monitor
In order to serve customers better with the limited network resources, QoS can monitor service flow of
specified user in ingress interface, which can adapt to the distributed network resources.
14.1.6
Interface speed limitation
Interface speed limitation is the speed limit based on interface which limits the total speed rate of interface
outputting packet.
14.1.7
Redirection
User can re-specify the packet transmission interface based on the need of its own QoS strategies.
14.1.8
Priority mark
Ethernet switch can provide priority mark service for specified packet, which includes : TOS, DSCP,
802.1p. These priority marks can adapt different QoS model and can be defined in these different models.
14.1.9
Choose interface outputting queue for packet
Ethernet switch can choose corresponding outputting queue for specified packets.
14.1.10
Queue scheduler
It adopts queue scheduler to solve the problem of resource contention of many packets when network
congestion. There are three queue scheduler matchings : Strict-Priority Queue (PQ), Weighted Round Robin (WRR)
and WRR with maximum delay.
14.1.10.1 PQ
PQ(Priority Queueing)is designed for key service application. Key service possesses an important feature,
that is, require the precesent service to reduce the response delay when network congestion. Priority queue divides all
packets into 4 levels, that is, superior priority, middle priority, normal priority and inferior priority (3, 2, 1, 0), and
their priority levels reduce in turn.
When queue schedulerimg, PQ precedently transmits the packets in superior priority according to the
priority level. Transmit packet in inferior priority when the superior one is empty. Put the key service in the superior
one, and non-key service (such as email)in inferior one to guarantee the packets in superior group can be first
transmitted and non-key service can be transmitted in the spare time.
The shortage of PQ is : when there is network congestion, there are more packets in superior group for a
long time, the packets in inferior priority will wait longer.
14.1.10.2 WRR
WRR queue scheduler divides a port into 4 or 8 outputting queues (QSW-3900 has 8 queues, that is, 3, 2, 1,
0) and each scheduler is in turn to guarantee the service time for each queue. WRR can configure a weighted value
(that is, w3, w2, w1, w0 in turn) which means the percentage of obtaining the resources. For example : There is a
port of 100M. Configure its WRR queue scheduler value to be 50, 30, 10, 10 (corresponding w3, w2, w1, w0 in turn)
to guarantee the inferior priority queue to gain at least 10Mbit/s bandwidth, to avoid the shartage of PQ queue
scheduler in which packets may not gain the service.
WRR possesses another advantage. The scheduler of many queues is in turn, but the time for service is not
fixed
——
if some queue is free, it will change to the next queue scheduler to make full use of bandwidth resources.
