Skip to content
RoCE course / Module 2

RoCE and RDMA for AI fabrics

0%

0 of 6 cases closed

Congestion

Build and defend the whole marking-and-congestion chain — application ToS through host trust state to switch queue, PFC vs ECN vs ZTR-RTT — and say which one a customer's symptom actually points at.

  1. 2.1 ToS, DSCP, PCP, SL: where marking gets lostThe capture that ends the argument
  2. 2.2 Host QoS with mlnx_qos: trust state, prio2tc, ETSTwo identical nodes that are not identical
  3. 2.3 PFC, lossless and the case for lossyThe architect who wants more pause
  4. 2.4 ECN, CNP and DCQCN: closing the loop02:10, and nothing is slowing down
  5. 2.5 Zero Touch RoCE, ZTR-RTT CC and DOCA PCCThe line item they refuse to buy
  6. 2.6 Packet spraying and the reordering it hands to the NICThe sixth word on the label
  7. Checkpoint