An Improved Budget-Deadline Constrained
Workflow Scheduling
Algorithm on Heterogeneous Resources
Ting Sun
∗
, Chuangbai Xiao
∗
, Xiujie Xu
∗
∗
College of Computer Science and Technology
Beijng University of Technology
Beijing, 100124, P.R.China
email: 971834529@qq.com
Guozhong Tian
†
†
Department of Computer Information and Engineering
Changzhou Institute of Technology
213002, P.R.China
Abstract—Abstract– In recent years, there are many schedul-
ing algorithms for execution of workflow applications using
Quality of Service (QoS) parameters. In this paper, we improve
a scheduling workflow algorithm considering the time and cost
constraints on heterogeneous resources, which is called Budget-
Deadline constrained using Sub-Deadline scheduling (BDSD).
With the deadline and budget constraints required by the user, we
use the BDSD algorithm to find a scheduling which satisfy with
the both constraints. We use the planning successful rate (PSR)
to show the effectiveness of our algorithm. In the simulation
experiment, we use the random workflow applications and real
workflow applications to experiment. The simulation results show
that compared with other algorithms, our BDSD algorithm has
a high PSR and low-time complexity of 𝑂(𝑛
2
𝑚) for 𝑛 tasks and
𝑚 processors.
Index Terms—DAG scheduling, workflow planning, sub-
deadline, quality of service, planning success rate.
I. INTRODUCTION
The workflow scheduling problem has been studied over
past years, focusing on heterogeneous computing systems
and distributed environments like grids and clusters. The
workflow scheduling problem is a form of task scheduling
problem. The challenges of the problem is: NP-hard nature of
task-resource mapping, diverse QoS requirements and mixed
resource scheduling. It used the directed acyclic graph(DAG)
to represent the workflow application where nodes represent
the application tasks and edges represent the inter-task data
dependencies. In heterogeneous computing systems, an effec-
tive scheduling is very important for executing applications.
Recently, workflow s cheduling to satisfy QoS parameter has
become activity in the utility computing. Yu et al.[1] pre-
sented some classical workflow scheduling algorithms. These
algorithms consider a single objective, such as minimising
makespan or minimising cost. However, with the user’s QoS
requirements adding, more and more algorithms considered
multiple QoS parameters. Some scheduling considered bi-
objective model that consists two QoS parameters, such as
time and cost at the same time, that problem becomes more
challenging[2]. The DAG scheduling problem has been proved
to be a NP-complete problem[3]. Furthermore, scheduling
tasks with uniform weights to any number of processors and
scheduling tasks with weights equal to one or two units to
two processors, which were also proved to be NP-complete
problems[4].
The two primary optimization problems: minimizing cost
under deadline constraint or minimizing makespan under bud-
get constraint have been extensively studied recently. Based on
available information of resources and workflows, it can divide
the workflow scheduling into different problems: best-effort
workflow scheduling, deadline-constrained workflow schedul-
ing, budget-constrained workflow scheduling, robust workflow
scheduling, multi-criteria workflow scheduling, data-intensive
workflow scheduling, energy-aware workflow scheduling, hy-
brid resource scheduling, cloud resource accessing and pricing
and so on[5]. Many papers also research the workflow schedul-
ing problem in the cloud environment. A cloud is a cluster
of distributed computers that provides on-demand computing
resources or services to the remote users over a network[6].
The cloud user can request multiple cloud services at the
same time. Cloud has the autonomic characteristics and the
VMs have the diversity characteristics. In Infrastructure-as-a-
Service (IaaS) cloud computing, computational resources are
provided for remote users in the form of leases. Li and Qiu
et al.[7] has proposed two online dynamic resource allocation
algorithms for the IaaS cloud system with preemptable tasks.
It adjust the resource allocation dynamically based on the
updated information of the actual task executions. Qiu et
al.[8] presented a genetic-based optimization algorithm for
chip multiprocessor in green clouds, which used the phase-
change memory (PCM) technique. It focuses on provide
a configuration that make PCM memory performance and
efficiency. Gai, Qiu et al.[9] in cloud computing given a cost-
aware multimedia data allocation for heterogeneous memory
using genetic algorithm. Xu et al.[10] used the slot extension
backfill strategy to make full use of the time slot between
tasks, but did not take into account the problem of scheduling
costs.
Many papers have been presented for multi-objective
scheduling [11], [12], [13], [15], [16], [17], [18], [19]. Zhang
2017 IEEE 4th International Conference on Cyber Security and Cloud Computing
978-1-5090-6644-5/17 $31.00 © 2017 IEEE
DOI 10.1109/CSCloud.2017.8
40