Information Technology Reference
In-Depth Information
Adaptive Spot-Instances Aware Autoscaling
for Scientific Workflows on the Cloud
David A. Monge
1
,
2
and Carlos Garcıa Garino
1
,
3
1
ITIC Research Institute, National University of Cuyo (UNCuyo), Argentina
2
Faculty of Exact and Natural Sciences, UNCuyo, Argentina
3
Faculty of Engineering, UNCuyo, Argentina
{
dmonge,cgarcia
}
@itu.uncu.edu.ar
Abstract.
This paper deals with the problem of autoscaling for cloud
computing scientific workflows. Autoscaling is a process in which the in-
frastructure scaling (i.e. determining the number and type of instances to
acquire for executing an application) interleaves with the scheduling of
tasks for reducing time and monetary cost of executions. This work pro-
poses a novel strategy called Spots Instances Aware Autoscaling (SIAA)
designed for the optimized execution of scientific workflow applications.
SIAA takes advantage of the better prices of Amazon's EC2-like spot
instances to achieve better performance and cost savings. To deal with
execution eciency, SIAA uses a novel heuristic scheduling algorithm to
optimize workflow makespan and reduce the effect of tasks failures that
may occur by the use of spot instances. Experiments were carried out us-
ing several types of real-world scientific workflows. Results demonstrated
that SIAA is able to greatly overcome the performance of state-of-the-art
autoscaling mechanisms in terms of makespan (up to 88.0%) and cost of
execution (up to 43.6%).
Keywords:
Scientific workflows, Cloud Computing, Autoscaling,
Scheduling, Spot instances.
1 Introduction
Many scientific areas have turn to in silico experimentation giving birth to the
so-called discipline
e-Science
. In this sense, workflow technology plays a central
role and has been widely adopted for guiding the design and execution of complex
scientific experiments [13]. Workflow applications comprise a set of computation
tasks and a set of dependencies between them, which determine constraints for
the execution order of tasks arranged in a directed acyclic graph (see figure 1).
To meet the computational requirements, which are usually high and eciently
execute the applications, cloud computing technologies are being extensively
used [17,11].
Public cloud providers permit a transparent, on-demand and inexpensive
access to computational resources relying on virtualization strategies [4]. In-
frastructure as a Service (IaaS) providers permit the
on-demand
acquisition
of Virtual Machine (VM)
instances
under a pay-per-use fashion with a fixed
