Workload Scheduler Efficiency Analysis
infolatencyUpdated Oct 24, 2022
Measure VM provisioning time from request to active state to identify scheduler bottlenecks, slow image pulls, or placement inefficiencies.
Sources
How to detect:
Track time between nova.compute.instance.create.start and nova.compute.instance.create.end notifications. Monitor nova-scheduler queue depth and current_workload per hypervisor. Alert when provisioning times exceed SLA thresholds.
Recommended action:
Analyze scheduler filters and weights for optimization opportunities. Check Glance image cache hit rates. Review network bandwidth for image transfers. Consider pre-caching frequently used images on compute nodes.