Enhanced Resilience and Scale Variations

Five Controller Deployments

Enhanced resiliency and performance for Contrail Networking and Openstack Red Hat can be achieved by running controller VMs on additional servers and providing Layer 2 connectivity between the controller VMs.

In this use case, enhanced resiliency is achieved by running controller VMs on five servers with Layer 2 connectivity.

The setup includes:

• 5 Contrail Controllers

• 5 Openstack Controllers

• 3 Analytic nodes

• 3 Analytics DB nodes

• 3 Appformix nodes

The enhanced resiliency provided by this architecture ensures the Contrail Cloud environments remains operational in cases where two controller nodes fail.

Controller VM placement in this architecture is configured in the site.yml file.

A configuration snippet from the site.yml file illustrating this controller VM placement:

Separated Controller Functions

In environments with large amounts of control plane or analytics traffic, you might want to increase scalability by splitting the OpenStack, Contrail Controller, and Contrail Analytics (with AppFormix) functions onto separate servers. This setup leads to base environments where 9 controller nodes are located in 3 racks.

Table 1 lists the roles for each controller node in separated controller environments.

When groups of software components are allocated to different control hosts, the networking requirement of each control host is the union of the requirements of the software roles running on the host. Specialized configurations can be configured using a different hardware profile for each combination of software components.