Alcatel-Lucent 7450 Quality Of Service Manual page 787

Scheduling Class Weighted Groups
As indicated above, an HSMDA scheduler handles groups of queues based on each queues
identifier. All queues numbered 1 are automatically placed in scheduler class 1, queues
numbered 2 are placed in scheduler class 2 through queues numbered 8 being placed in
scheduler class 8. Each scheduling class may be directly associated with a strict priority level
or may be placed in one of two weighted groups. Each weighted group is used to map up to
three scheduling classes into a single strict priority level and provides a weight for each
member class of the group. Using the weighted groups allows for a mixture of strict and
weighted scheduling between the scheduling classes. The scheduling classes mapped to a
group must be consecutive in class order. If class 3 is placed into group 1, then the next class
that may be placed into the group could be 2 or 4. If 4 were added to the group, then the next
(and last) class that can be added to the group would be 5. If 2 had been added to the group
instead of 4, then the next class would be limited to class 1.
Scheduler Strict Priority Levels
The scheduler maintains 8 strict levels with strict level 8 being the highest priority and strict
level 1 being the lowest. For each strict priority level, either the scheduler class with the same
ID (1 through 8) may be mapped to the strict level, or one of the two weighted groups may
be mapped to the strict level. If a scheduling class is not mapped to a weighted group, the class
is instead mapped to its relative strict scheduling level. Once a scheduling class is mapped to
a weighted group, it is removed from the strict level and shares the strict level assigned to the
weighted group. The weighted group itself is mapped to the inherent strict scheduling level
of the highest member scheduling class. It should be apparent that when weighted scheduling
class groups are used, fewer strict levels are active on the scheduler.
Strict Priority Level PIR
The scheduler supports a strict scheduling level PIR that limits the amount of bandwidth
allowed for the level. The rate is defined in increments of megabits per second and may be
set to max (the default setting) which disables the shaping function. The shaping rate is not
defined on the strict priority level, but is inherited from the scheduling class or weighted
group that is mapped to the strict level. The scheduler includes the full Ethernet frame
encapsulation overhead when updating the priority level PIR, including the 12 byte inter-
frame gap and the 8 byte preamble.
Quality of Service Guide
High Scale Ethernet MDA Capabilities
789