User manual

C HAPTER 11 | Class of Service
Layer 2 Queue Settings
– 127 –
ES-2000 Series
The basic WRR algorithm uses a relative weight for each queue that
determines the percentage of service time the switch services each
queue before moving on to the next queue. This prevents the head-of-
line blocking that can occur with strict priority queuing.
Deficit Weighted Round-Robin (DWRR) services the queues in a manner
similar to WRR, but the next queue is serviced only when the queue’s
Deficit Counter becomes smaller than the packet size to be transmitted.
As a result, traffic on queues with large weights cause increased latency
and jitter for traffic waiting for scheduling other queues. In SDWRR, if
two or more queues have traffic eligible for transmission (i.e. the Deficit
Counter is greater than the packet size to be transmitted), then a
round-robin scheme among those queues is used, while still preserving
the overall weight ratios between the queues. This produces less jitter
and lower maximum latency for traffic on all of the serviced queues.
◆ If Strict and WRR mode is selected, a combination of strict service is
used for the high priority queues and weighted service for the
remaining queues. The queues assigned to use strict priority should be
specified using the Strict Mode field parameter.
◆ A weight can be assigned to each of the weighted queues (and thereby
to the corresponding traffic priorities). This weight sets the frequency
at which each queue is polled for service, and subsequently affects the
response time for software applications assigned a specific priority
value.
Service time is shared at the egress ports by defining scheduling
weights for SWDRR, or the queuing mode that uses a combination of
strict and weighted queuing. Service time is allocated to each queue by
calculating a precise number of bytes per second that will be serviced
on each round.
◆ The specified queue mode applies to all interfaces.
PARAMETERS
These parameters are displayed:
◆ Queue Mode
■
Strict – Services the egress queues in sequential order,
transmitting all traffic in the higher priority queues before servicing
lower priority queues. This ensures that the highest priority packets
are always serviced first, ahead of all other traffic.
■
WRR (SWDRR) – Shares bandwidth at the egress ports by using
scheduling weights, servicing each queue in a round-robin fashion.
■
Strict and WRR – Uses strict priority on the high-priority queues
and SDWRR for the rest of the queues. (This is the default setting.)
◆ Queue ID – The ID of the priority queue. (Range: 0-7)