MSc Defense: Monitoring Elastically Adaptive Cloud Services, 18 September 2014
Demetris Trihinas is going to defense his master thesis entitled: "Monitoring Elastically Adaptive Cloud Services", on 18 September 2014. The MSc. defense will take place on Room 148, Building 12 Faculty of Pure and Applied Sciences, New Campus, at 10-12 am.
Over the past decade, cloud computing has rapidly become a widely accepted paradigm with core concepts such as elasticity, scalability and on-demand automatic resource provisioning emerging as next generation cloud service -must have- properties. Elastic cloud services have the ability to expand or contract the resources dedicated to them in order to meet the current demand. However, automatically provisioning resources for elastic cloud services is not a trivial task, requiring for both the applications and the underlying platform(s) to be constantly monitored at various levels and time intervals. The complex nature of this task requires for the monitoring system to automatically detect configuration changes in the application topology due to elasticity actions or when allocated resource-related parameters change. Existing monitoring solutions either leave this challenge unaddressed or sacrifice their portability and interoperability in favor of elasticity support. In this talk, we discuss what are the challenges which arise when monitoring elastic cloud services which can even be distributed across multiple cloud platforms. To address these challenges, we introduce JCatascopia; a fully automated, multi-layer, interoperable cloud monitoring system. JCatascopia is able to run in a non-intrusive and transparent manner to any underlying virtualized infrastructure since neither the metric collection process nor metric distribution are depended to the underlying platform. In contrast to the other solutions, JCatascopia uses a variation of the publish and subscribe message pattern to dynamically detect configuration changes to the overall system due to elasticity actions without limiting its portability. Experiments on different cloud platforms show that JCatascopia is capable of supporting automated resource provisioners with proven interoperability, scalability and low runtime footprint.
Demetris Trihinas is a Master's candidate at the Department of Computer Science of the University of Cyprus and is also a Research Assistant at the Laboratory for Internet Computing (LInC). He holds a Dipl.-Ing. degree in Electrical and Computer Engineering from the National Technical University of Athens (NTUA) since 2012. His research interests include Distributed Systems and Architectures, Web Services, Monitoring and Data Analysis, Cloud Computing, and Information Retrieval from Grid and Cloud Infrastructures. He is currently contributing as a researcher to the EU FP7 co-funded project named CELAR (Cloud ELAsticity pRovisioning) that aims in developing a fully automated multi-grained platform for elastic resource provisioning for cloud applications.