Red Hat Bugzilla – Bug 1280475
High latency seen in Neutron hops during performance tests
Last modified: 2016-04-26 12:29:03 EDT
This environment has two regions. On one of them, high latency is being seen in the hops within Openstack/neutron and then cleaning up once within the core network. The spikes occur when performance tests are run at certain intervals. The tests are run in both regions, and the issues are only seen within the one region. All instances see this pattern.
The network layout is several UCS chassis (blades) for our neutron/mgmt nodes connected up to an Fabric Interconnect, and then off to the Dist layer of Cisco 5k's and up and through the Core. 2x 10Gb connections to the FI coming out of Openstack blades.