Priority-based scheduling
Priority-based scheduling is a policy that selects or orders eligible work for execution according to assigned relative priorities.
Distributed Machine Learning Patterns
Priority scheduling is a scheduling algorithm that orders work according to assigned priority levels and may preempt lower-priority work.
- How it is used
- Prioritizes shared-cluster jobs so urgent machine-learning work can proceed despite limited compute capacity.
- How deep it goes
- Defines priority ordering and preemption through queue and running-job scenarios, then addresses governance risks without configuration details.
- Examples
- Illustrates four jobs ranked high, medium, and low, plus preemption of running low-priority job B by high-priority job A.
- What you can do after
- Evaluate/understand when priority ordering and preemption unblock urgent work, and when priority limits are needed to protect other jobs.
Read how Distributed Machine Learning Patterns explains priority-based scheduling →
Kubernetes for Developers
Priority is an integer value assigned to a Pod that influences its scheduling order and eviction ranking.
- How it is used
- Ranks Pods for scheduling and eviction when cluster resources are constrained.
- How deep it goes
- Defines integer priorities and ranking behavior, then demonstrates PriorityClass and Deployment configuration.
- Examples
- Uses high-priority YAML, a three-replica timeserver Deployment, and alternative preempting and non-preempting policies.
- What you can do after
- Configure Pod priority with PriorityClass and priorityClassName, including whether lower-priority Pods may be preempted.
Read how Kubernetes for Developers explains priority-based scheduling →
Related concepts
- fair-share schedulingcontrasts with · Fair-share scheduling is an operating-system scheduling algorithm that distributes CPU usage among users or groups rather than equally among individual processes.
- starvationaddresses · Starvation is a condition in which a task persistently lacks the resources or execution opportunities needed to make progress while competing work proceeds.
- preemptionuses · Preemption is the interruption or displacement of active execution or resource allocation so another eligible activity can proceed.
- scheduleruses · A scheduler is a system component that selects, orders, dispatches, or assigns execution resources to tasks according to timing, dependency, placement, or other scheduling criteria.
- deadlockaddresses · A deadlock is a state in which entities cannot make progress because each depends on a resource, event, or action blocked by another.
- PriorityClassconfigured by · A PriorityClass is a Kubernetes resource defining a pod scheduling priority and associated preemption behavior.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.