Skip to main content

Table 1 QoS-driven scheduler characterization

From: QoS-driven scheduling in the cloud

Feature

Classification

Workload source

Multi-users and multi-job

Job Structure

Single-task

Job flexibility

Rigid

Arrival process

Open

Workload composition

Heterogeneous

Quality of service

SLO aware

Real time

No real time

Resources heterogeneity

Any

Scaling

Static

Resource sharing

Dedicated VMs/containers

Geographical coverage

Local

Federation

Single domain

Scheduling goal

SLO accomplishment

Level

Task-level

Data locality

No affinity

Failure model

Crash-recovery

Adaptability

Static

Optimality

Sub-optimal

Operation

Online

Topology

Distributed, centralized, push-based

Flexibility

Flexible: migration-aware, preemptive