Question 1
A new data center connects servers on separate leaf switches and an engineer expects those racks to traverse an unrelated access switch rather than the common spine tier. Which explanation BEST identifies the Juniper data-center or Apstra mechanism causing this result?
Explanation
Correct Answer
In a conventional leaf–spine Clos fabric, interleaf traffic normally crosses an available spine while servers attach at the leaf edge. A test flow between leaf-attached hosts follows a supported leaf–spine–leaf path and avoids an unintended transit leaf.
Incorrect Answers
A distributed anycast gateway design requires a consistent logical gateway identity across participating VTEPs for the same tenant segment. That concerns anycast default gateway identity across leafs, while the issue here is leaf–spine predictable hop count. Trace the actual leaf-to-spine adjacencies and restore the reference topology so each leaf has its intended uplinks to the spine set.
Edge-routed bridging places an appropriate distributed Layer 3 gateway function at fabric edges for attached virtual networks. However, the present diagnosis decision is grounded in: A new data center connects servers on separate leaf switches and an engineer expects those racks to traverse an unrelated access switch rather than the common spine tier. For this issue, trace the actual leaf-to-spine adjacencies and restore the reference topology so each leaf has its intended uplinks to the spine set.
EVPN control-plane MAC mobility handling updates the advertised endpoint location and withdraws or supersedes stale location information. Here, in a conventional leaf–spine Clos fabric, interleaf traffic normally crosses an available spine while servers attach at the leaf edge.