Two-Tier Scheduling Strategy

A Two-Tier Scheduling Strategy is a distributed scheduling strategy that can support two-tier scheduling tasks.

• AKA: Two-Level Scheduling, Offer-Based Scheduling, Pessimistic Concurrency Control Scheduling, Mesos-Style Scheduling.
• Context:
  • It can typically separate Resource Allocation Decisions through central resource managers and framework-specific schedulers.
  • It can typically implement Resource Offer Mechanisms using resource offer protocols and resource acceptance filters.
  • It can typically manage Scheduling Autonomy via framework scheduler independence and framework-specific optimizations.
  • It can typically provide Resource Fairness through dominant resource fairness algorithms and resource share guarantees.
  • It can typically handle Resource Rejections using offer decline reasons and resource re-offer cycles.
  • It can often enable Framework Coexistence via resource partitioning schemes and framework priority levels.
  • It can often support Resource Revocation through revocable resource offers and preemption protocols.
  • It can often facilitate Scheduling Efficiency using resource offer filters and allocation modules.
  • It can range from being a Simple Two-Tier Scheduling Strategy to being a Complex Two-Tier Scheduling Strategy, depending on its scheduling policy sophistication.
  • It can range from being a Homogeneous Two-Tier Scheduling Strategy to being a Heterogeneous Two-Tier Scheduling Strategy, depending on its resource type diversity.
  • It can range from being a Static Two-Tier Scheduling Strategy to being a Dynamic Two-Tier Scheduling Strategy, depending on its offer pattern adaptability.
  • It can range from being a Pessimistic Two-Tier Scheduling Strategy to being a Optimistic Two-Tier Scheduling Strategy, depending on its concurrency control approach.
  • ...
• Examples:
  • Production Two-Tier Scheduling Implementations, such as:
    • Apache Mesos Two-Tier Scheduling for datacenter two-tier scheduling.
    • Google Omega Scheduling for shared-state two-tier scheduling.
  • Two-Tier Scheduling Frameworks using this strategy, such as:
    • Marathon on Mesos for container two-tier scheduling.
    • Apache Spark on Mesos for data processing two-tier scheduling.
    • Chronos on Mesos for job two-tier scheduling.
  • Academic Two-Tier Scheduling Research, such as:
    • DRF Algorithm Implementations for fairness-focused two-tier scheduling.
    • Offer-Based Scheduling Studys for efficiency-focused two-tier scheduling.
  • ...
• Counter-Examples:
  • Monolithic Scheduling Strategy, which uses centralized scheduler rather than two-tier architecture.
  • Shared-State Scheduling Strategy, which provides optimistic concurrency rather than pessimistic offer-based control.
  • Fully Distributed Scheduling Strategy, which lacks central resource manager component.
• See: Distributed Scheduling Strategy, Resource Scheduling Algorithm, Apache Mesos Distributed Resource Management Platform, Scheduling Architecture Pattern, Resource Offer Protocol, Dominant Resource Fairness, Cluster Scheduling System.

---