os-nosdn-performance-ha overview and description

This document provides scenario level details for Danube 1.0 of deployment with no SDN controller and performance options enabled.

1. Introduction

This scenario is used primarily to demonstrate the performance settings and capabilities in Apex. This scenario will deploy a Newton OpenStack deployment without any NFV features or SDN controller enabled.

1.1. Scenario components and composition

This scenario is composed of common OpenStack services enabled by default, including Nova, Neutron, Glance, Cinder, Keystone, Horizon. Optionally and by default, Tacker and Congress services are also enabled. Ceph is used as the backend storage to Cinder on all deployed nodes.

All services are in HA, meaning that there are multiple cloned instances of each service, and they are balanced by HA Proxy using a Virtual IP Address per service.

The main purpose of this scenario is to serve as an example to show how to set optional performance settings in an Apex deploy settings file.

1.2. Scenario usage overview

The performance options listed in os-nosdn-performance-ha.yaml give an example of the different options a user can set in any deploy settings file. Some of these performance options are actually required for other scenarios which rely on DPDK. Options under the nova section like ‘libvirtpin’ allow a user to choose which core to pin nova instances to on the overcloud compute node. Options under ‘kernel’ allow a user to set kernel specific arguments at boot, which include options like hugepages, isolcpus, enabling iommu, etc.

1.3. Limitations, Issues and Workarounds

  • APEX-389:
    Compute kernel parameters are applied to all nodes

1.4. References

For more information on the OPNFV Danube release, please visit http://www.opnfv.org/danube