JCloudScale: Closing the Gap Between IaaS and PaaS
Authors:
Rostyslav Zabolotnyi,
Philipp Leitner,
Waldemar Hummer,
Schahram Dustdar
Abstract:
The Infrastructure-as-a-Service (IaaS) model of cloud computing is a promising approach towards building elastically scaling systems. Unfortunately, building such applications today is a complex, repetitive and error-prone endeavor, as IaaS does not provide any abstraction on top of naked virtual machines. Hence, all functionality related to elasticity needs to be implemented anew for each applica…
▽ More
The Infrastructure-as-a-Service (IaaS) model of cloud computing is a promising approach towards building elastically scaling systems. Unfortunately, building such applications today is a complex, repetitive and error-prone endeavor, as IaaS does not provide any abstraction on top of naked virtual machines. Hence, all functionality related to elasticity needs to be implemented anew for each application. In this paper, we present JCloudScale, a Java-based middleware that supports building elastic applications on top of a public or private IaaS cloud. JCloudScale allows to easily bring applications to the cloud, with minimal changes to the application code. We discuss the general architecture of the middleware as well as its technical features, and evaluate our system with regard to both, user acceptance (based on a user study) and performance overhead. Our results indicate that JCloudScale indeed allowed many participants to build IaaS applications more efficiently, comparable to the convenience features provided by industrial Platform-as-a-Service (PaaS) solutions. However, unlike PaaS, using JCloudScale does not lead to a loss of control and vendor lock-in for the developer.
△ Less
Submitted 10 November, 2014;
originally announced November 2014.
