Class of Service (RDMA/RoCE) JN0-683 Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   Which Junos operational command is used to verify the PFC configuration and status on an interface?

   • Ashow cos classifiers
   • Bshow interfaces pfc
   • Cshow interfaces queue
   • Dshow interfaces extensive
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A Juniper QFX10002 switch is configured for RoCEv2. The administrator has configured a classifier to map DSCP 26 to forwarding class 'roce' and a scheduler with a transmit-rate of 'percent 100' and a buffer-size of 'percent 50'. PFC is enabled on the 'roce' queue. During a test, the 'show interfaces queue' output shows that the 'roce' queue is experiencing drops. What is the most likely cause?

   • APFC is not enabled on the ingress interface, so the switch cannot send pause frames to the upstream device.
   • BThe scheduler's transmit-rate is set to 100%, which prevents the queue from ever being paused, causing drops.
   • CThe buffer-size of 50% is insufficient for the bursty RoCE traffic, causing the queue to overflow before PFC can take effect.
   • DThe classifier is not mapping the DSCP value correctly, so the RoCE traffic is not being placed in the 'roce' queue.
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A Juniper QFX5110 switch is experiencing performance issues with RoCEv2 traffic. The 'show interfaces queue' output shows that the 'roce' queue is dropping packets. The administrator has verified that PFC is enabled on the 'roce' queue and that the buffer-size is adequate. What is the next step to troubleshoot the issue?

   • ACheck the 'show interfaces priority-flow-control' output to see if pause frames are being sent and received on the interface.
   • BDisable ECN on the 'roce' queue to see if ECN is causing the drops.
   • CReboot the switch to clear any potential software issues.
   • DIncrease the buffer-size for the 'roce' queue to 100% to ensure no drops occur.
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A data center is expanding its storage network to span multiple racks connected by a spine-leaf fabric. The storage traffic uses RDMA, and the current RoCEv1 deployment is limited to a single rack. The team needs to extend RDMA across the fabric. What is the primary reason RoCEv1 cannot be used, and what is the solution?

   • ARoCEv1 uses a different DSCP value than RoCEv2, causing classification issues; the solution is to update the classifier.
   • BRoCEv1 requires a lossless network, which is not possible in a spine-leaf fabric; the solution is to use TCP instead.
   • CRoCEv1 lacks the UDP header required for IP routing; the solution is to migrate to RoCEv2.
   • DRoCEv1 does not support PFC; the solution is to enable PFC on the spine and leaf switches.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   Which encapsulation does RoCEv2 use to enable routing across IP networks?

   • AUDP
   • BVXLAN
   • CGRE
   • DTCP
   Reveal Solution Discussion Select an answer first