원문정보
초록
영어
Cloud computing can provide the dynamic and elastic virtual resources for the users to execute the large-scale computing tasks. It has become the hot spot in the academic and industry fields. Task scheduling is one of the most important issues in the Cloud. In the Cloud systems, the goal of the tasks scheduling is to spread the workload among the computing nodes and maximize the utilization while the total execution time is within the specific delay bound. At present, almost scheduling algorithms focus on the single task dispatch in the Cloud. Unfortunately, there is little research on the associate tasks scheduling considering the deadline bound. In this paper, two hierarchical task models were discussed and the corresponding associated task scheduling algorithms based on delay-bound constraint (ATS-DB and SAH-DB) were proposed. The associated tasks and the task execution order were represented by one directed acyclic graph (DAG). The proposed hierarchical task models can improve the task execution concurrency. Extensive experimental results demonstrated that the proposed scheduling algorithms, ATS-DB and SAH-DB, can reduce the execution cost and improve the resource utilization within the user-expected delay bound.
목차
1. Introduction
2. Related Work
2.1. Independent Tasks Scheduling
2.2. Associated Task Scheduling
3. Task Hierarchical Model
3.1. Tasks DAG Model
3.2. Problem Statement
3.3. Example
4. CPM Scheduling Alogrithm
5. ATS-DB Scheduling Algorithm
5.1. Hierarchical Decomposition Method
5.2. Calculation Delay Bound
5.3. Detail of ATS-DB Algorithm
5.4. Example of ATS-DB Scheduling
5.5. Analysis of ATS-DB Algorithm
6. SAH-DB Scheduling Algorithm
6.1. SAH-based Hierarchical Decomposition Method
6.2. Detail of SAH-DB Algorithm
6.3. Example of SAH-DB Scheduling
7. Performance Evaluation
7.1. Experiments Settings & Methodology
7.2. Execution Cost with the Same Deadline
7.3. Optimization Ratio with Different Deadline
8. Conclusion and Future Work
References