Allied Telesis AT-S63 - Manual (Page 136)

Chapter 12: Class of Service
136 Section II: Advanced Operations
Scheduling
A switch port needs a mechanism for knowing the order in which it should
handle the packets in its eight egress queues. For example, if all the
queues contain packets, should the port transmit all packets from Q7, the
highest priority queue, before moving on to the other queues, or should it
instead just do a few packets from each queue and, if so, how many?
This control mechanism is called scheduling. Scheduling determines the
order in which a port handles the packets in its egress queues. The
AT-S63 software has two types of scheduling:
ˆ Strict priority
ˆ Weighted round robin priority
Note
Scheduling is set at the switch level. You cannot set this on a per-
port basis.
Strict Priority
Scheduling
With this type of scheduling, a port transmits all packets out of higher
priority queues before transmitting any from the lower priority queues. For
instance, as long as there are packets in Q7 it does not handle any
packets in Q6.
The value to this type of scheduling is that high priority packets are always
handled before low priority packets.
The problem with this method is that some low priority packets might
never be transmitted out the port because a port might never get to the low
priority queues. A port handling a large volume of high priority traffic may
be so busy transmitting that traffic that it never has an opportunity to get to
any of the packets stored in its low priority queues.
Weighted Round
Robin Priority
Scheduling
The weighted round robin scheduling method functions as its name
implies. The port transmits a set number of packets from each queue, in a
round robin fashion, so that each has a chance to transmit traffic. This
method guarantees that every queue receives some attention from the
port for transmitting packets.
To use this scheduling method, you need to specify the maximum number
of packets a port should transmit from a queue before moving to the next
queue. This is referred to as specifying the “weight” of a queue. In most
cases, you will want to give greater weight to the higher priority queues
over the lower priority queues.