Description of problem: Comparing OSP 16 with OSP 13z9, we see that API response times for listing ports is higher with OSP 16 than OSP 13. As a note OVN is being used as the ML2 backend in both cases. Scenario run: Create 4 ports per network and list all ports for a total of 1000 times at a concurrency of 16. As the number of ports to list increases, the time taken to list increases linearly. Rally scenario used: https://github.com/cloud-bulldozer/browbeat/blob/master/rally/neutron/neutron-create-list-port-cc.yml I have run several tests repeatedly to make sure this is not a one off. Max 95% Average Min OSP 16 43.381 22.68 11.96 0.241 OSP 13 14.759 9.842 5.205 0.159 Version-Release number of selected component (if applicable): 16 How reproducible: 100% Steps to Reproduce: 1. Run rally scenario to create-list-ports 2. Observe results 3. Compare OSP 13 and OSP 16 Actual results: Performance with OSp 16 is much worse than OSP 13 Expected results: Performance should be better than or equivalent to OSP 13 Additional info:
As a an extra data point, listing ports is signifcantly slower in OSP 16 when compared to OSP 13 even when using ML2/OVS.
Nate, any update on this?
Thanks Rodolfo. We will retest and publish the results once the puddle with fixes available.
Closing EOL, OSP 16.0 has been retired as of Oct 27, 2020
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Red Hat OpenStack Platform 16.1.3 bug fix and enhancement advisory), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2020:5413
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 500 days