OpenStack workload monitoring

The next remaining part of the operational task of cloud operators is to provide monitoring support to the running workloads of the OpenStack tenants. Operators should also monitor some generic metrics of the provisioned resources by tenants to anticipate quota limitations and gain insightful analysis of the infrastructure usage. This can be achieved by using the same toolsets covered in the previous section. Workload monitoring focuses more on the resources deployed by each tenant – more precisely, the Nova instances. Prometheus is capable of querying the Nova API, and along with its labeling configuration settings, you should be able to filter and organize a list of thousands of instances into a set of scraped targets. In the previous section, the Prometheus Libvirt exporter was deployed in each compute node (see Step 8 of the Deploying infrastructure monitoring section). Thus, there is no need to reconfigure any part of the infrastructure hosts...