White Papers

This default behavior is impacted if you modify the total buffer available for PFC or assign static buffer configurations to the
individual PFC queues.
Shared headroom for lossless or PFC packets
In switches that require lossless frame delivery, some fixed buffer is set aside to absorb any bursty traffic that arrives after
flow control is configured (PFC in this case). This extra buffer space is called the PG headroom. The additional buffer space is
reserved for ingress ports per PG. As the buffer is reserved per ingress Port and per PG, the total reserved headroom buffer is
the sum of the PG headroom buffer reserved for all PGs configured across all ingress ports on the switch.
The PG headroom allocation is done conservatively to guarantee lossless operation in worst case scenarios where huge amounts
of bursty traffic arrive at the ingress ports. However, this scheme of allocating headroom buffer per PG and per ingress port
may result in the wastage of the reserved PG headroom buffer; as, this headroom buffer may never be utilized and some of the
buffer space allocated to PG headroom is wasted.
To address this issue, Dell EMC Networking OS enables you to configure the shared headroom buffer for the entire device.
Each PG can utilize up to the peak headroom configured per PG as part of the buffer threshold profile. The traditional threshold
for any inflight or bursty traffic is set per ingress port and per PG. Retaining the same ingress admission control capabilities,
headroom pool can also be used to manage the headroom buffer as a shared resource.
Each PG can use the shared headroom pool only up to its PG headroom limit. The shared headroom feature provides the
capability to share the headroom buffer between all the ingress ports or PGs. It also provides ways to learn statistical data on
shared buffer usage, thereby, reducing the overall headroom buffer allocation.
The PFC Shared Headroom feature provides the following two capabilities:
● Headroom Pool Management – Provides the capability to use the shared headroom buffer between all the ingress ports or
PGs to reduce the overall headroom buffer allocation.
● Headroom Pool Monitoring – Provides a mechanism to monitor the peak headroom buffer consumed over a period of time,
which in turn helps in configuring a proper value for the shared headroom buffer.
This feature also provides a mechanism to monitor the peak headroom buffer consumed over a period of time, which in turn
helps you to configure a proper value for shared headroom buffer.
Example Scenario
Consider a scenario where you want to configure two lossless queues on 12 40 Gigabit ports.
The following table illustrates the buffer usage statistics when shared headroom is not used and each queue is allocated with a
fixed headroom buffer space:
Table 21. Buffer usage statistics when shared headroom is not used
Parameter Description
PFC Shared Buffer Size 208 KB
Buffer-size parameter of dcb-buffer-threshold
per lossless queue
94KB
Pause-threshold parameter of dcb-buffer-
threshold per lossless queue
18KB
The Headroom value reserved per lossless
queue
94 – 18 = 76KB
NOTE: 76KB is the headroom space that is required per PG [or a
lossless queue] on a 40 Gigabit port in a worst case scenario to
guarantee lossless behavior.
Total buffer required for 2 lossless queues on 12
ports
12*2*94 + 208 = 2464KB
NOTE: Out of the 2464KB, 1824KB [12*2*76KB] is reserved only for
the headroom, which might never be utilized in most of the cases. An
ideal approach is to increase the PFC shared buffer and reduce the
buffer reserved for headroom. However, this approach subsequently
reduces the time to trigger PFC, thereby, reducing the effectiveness
of PFC.
Data Center Bridging (DCB) 257