Expanding the compute farm

Managing the compute fleet hypervisors through kolla-ansible to spawn instances or containers is straightforward. After collecting your compute requirements, hardware capacity, types of workloads, and scheduling strategies, adding a compute node can be performed on the fly without incurring a complex operation overhead. The main requirements that should be considered are the listing of compute services – mainly, nova-compute and a networking plugin agent to provide connectivity with the control plane and sync with nova-scheduler for each resource allocation. The previous section demonstrated that additional components can be part of the compute node, including the Zun compute service, if you’re planning to offer an easy and quick container playground to end users. Another aspect of managing compute nodes from code through kolla-ansible is listing the available configuration options of the compute service, such as overcommitment ratios for CPU...