I will defer to the Wisconsin and Clemson admins for the final word on
this, but this is my understanding of the switch topologies.
The Wisconsin d7525 nodes are all connected to the same Dell Z9432F
switch. Unfortunately this switch is the core switch for the experiment
network, not only does it have a number of nodes directly connected,
but also is the aggregation point for four other switches and the dataset
storage server. So it is theoretically possible that traffic from all other
experiments could pass through this switch. That said, since the experiment
resource mapping tool favors intraswitch connections whenever possible, most
experiments will have their traffic confined to a single switch, so that
unless they are using multiple nodes types in their experiment or they are
using a remote dataset, their traffic will not traverse the Z9432.
The Clemson r7525 nodes are all connected to a Dell Z9264F switch.
I believe those are the only nodes/switches connected to that switch,
but traffic from other experiments using r7525 nodes will traverse this
switch.
> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to
cloudlab-user...@googlegroups.com.
> To view this discussion on the web visit
https://groups.google.com/d/msgid/
> cloudlab-users/9e2cb569-5076-4f3a-83df-224775590aa7n%
40googlegroups.com.