We were talking about the stress test:
Step 1 measure dataplane performance between a pair of VMs
(baseline)
Step 2 Begin the lifecycle of a new VM pair and
measure dataplane performance between a pair of VMs
Step 3 (Under discussion)
Measure traffic (throughput, loss)
between the second pair of VMs at the same time
as the first pair.
When you do this, you should see:
1. reduced ability to generate traffic and receive
traffic correctly in each VM pair (under stress)
2. reduced throughput and increased loss from the
the single VM-pair case (baseline).
3. the change from the baseline may be
more than a factor of 2 (reduction for Thrpt,
increase for Loss)
I think this form of stress test would illustrate
the deviation from linear when increasing workload.
However, it may only be applicable to the test VM-pairs,
and not extend to VNFs in general (although that would be
another aspect to test.
HNY everybody,
Al
_______________________________________________
opnfv-tech-discuss mailing list
[email protected]
https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
