Create and Import a DCQCN Configlet
For instructions about how to create and import a DCQCN Configlet into your Apstra blueprint, see Enable DCQCN in your AI Fabric.
For more information about DCQCN, see Data Center Quantized Congestion Notification (DCQCN).
The following is an example DCQCN Configlet:
class-of-service {
classifiers {
dscp mydscp {
forwarding-class CNP {
loss-priority low code-points 110000;
}
forwarding-class NO-LOSS {
loss-priority low code-points 011010;
}
}
}
drop-profiles {
dp1 {
interpolate {
fill-level [ 55 90 ];
drop-probability [ 0 100 ];
}
}
}
shared-buffer {
ingress {
buffer-partition lossless {
percent 66;
dynamic-threshold 10;
}
buffer-partition lossless-headroom {
percent 24;
}
buffer-partition lossy {
percent 10;
}
}
egress {
buffer-partition lossless {
percent 66;
}
buffer-partition lossy {
percent 10;
}
}
}
dedicated-buffer {
ingress {
percent 15;
}
egress {
percent 30;
}
}
forwarding-classes {
class CNP queue-num 3;
class NO-LOSS queue-num 4 no-loss pfc-priority 3;
}
congestion-notification-profile {
cnp {
pfc-watchdog;
input {
dscp {
code-point 011010 {
pfc;
}
}
}
output {
ieee-802.1 {
code-point 011 {
flow-control-queue 4;
}
}
}
}
}
interfaces {
et-* {
congestion-notification-profile cnp;
scheduler-map sm1;
unit * {
classifiers {
dscp mydscp;
}
}
}
}
scheduler-maps {
sm1 {
forwarding-class CNP scheduler s2-cnp;
forwarding-class NO-LOSS scheduler s1;
}
}
schedulers {
s1 {
drop-profile-map loss-priority any protocol any drop-profile dp1;
explicit-congestion-notification;
}
s2-cnp {
transmit-rate percent 5;
priority strict-high;
}
}
}After you import and commit, Apstra includes the DCQCN policies in the
config-context for the relevant switches in your fabric.
ConnectorOps RNIC Configurator automatically reads these policies and enforces them on
all GPU server RNICs.